Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
/*
|
|
|
|
* LibXDiff by Davide Libenzi ( File Differential Library )
|
|
|
|
* Copyright (C) 2003 Davide Libenzi
|
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
2017-11-07 05:39:33 +00:00
|
|
|
* License along with this library; if not, see
|
|
|
|
* <http://www.gnu.org/licenses/>.
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
*
|
|
|
|
* Davide Libenzi <davidel@xmailserver.org>
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "xinclude.h"
|
|
|
|
|
|
|
|
static long xdl_get_rec(xdfile_t *xdf, long ri, char const **rec) {
|
|
|
|
|
|
|
|
*rec = xdf->recs[ri]->ptr;
|
|
|
|
|
|
|
|
return xdf->recs[ri]->size;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int xdl_emit_record(xdfile_t *xdf, long ri, char const *pre, xdemitcb_t *ecb) {
|
|
|
|
long size, psize = strlen(pre);
|
|
|
|
char const *rec;
|
|
|
|
|
|
|
|
size = xdl_get_rec(xdf, ri, &rec);
|
|
|
|
if (xdl_emit_diffrec(rec, size, pre, psize, ecb) < 0) {
|
|
|
|
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Starting at the passed change atom, find the latest change atom to be included
|
|
|
|
* inside the differential hunk according to the specified configuration.
|
2013-06-19 18:46:07 +00:00
|
|
|
* Also advance xscr if the first changes must be discarded.
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
*/
|
2013-06-19 18:46:07 +00:00
|
|
|
xdchange_t *xdl_get_hunk(xdchange_t **xscr, xdemitconf_t const *xecfg)
|
|
|
|
{
|
|
|
|
xdchange_t *xch, *xchp, *lxch;
|
2008-12-28 18:45:32 +00:00
|
|
|
long max_common = 2 * xecfg->ctxlen + xecfg->interhunkctxlen;
|
2013-06-19 18:46:07 +00:00
|
|
|
long max_ignorable = xecfg->ctxlen;
|
|
|
|
unsigned long ignored = 0; /* number of ignored blank lines */
|
|
|
|
|
|
|
|
/* remove ignorable changes that are too far before other changes */
|
|
|
|
for (xchp = *xscr; xchp && xchp->ignore; xchp = xchp->next) {
|
|
|
|
xch = xchp->next;
|
|
|
|
|
|
|
|
if (xch == NULL ||
|
|
|
|
xch->i1 - (xchp->i1 + xchp->chg1) >= max_ignorable)
|
|
|
|
*xscr = xch;
|
|
|
|
}
|
|
|
|
|
2022-05-02 16:50:37 +00:00
|
|
|
if (!*xscr)
|
2013-06-19 18:46:07 +00:00
|
|
|
return NULL;
|
|
|
|
|
|
|
|
lxch = *xscr;
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
|
2013-06-19 18:46:07 +00:00
|
|
|
for (xchp = *xscr, xch = xchp->next; xch; xchp = xch, xch = xch->next) {
|
|
|
|
long distance = xch->i1 - (xchp->i1 + xchp->chg1);
|
|
|
|
if (distance > max_common)
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
break;
|
|
|
|
|
2013-06-19 18:46:07 +00:00
|
|
|
if (distance < max_ignorable && (!xch->ignore || lxch == xchp)) {
|
|
|
|
lxch = xch;
|
|
|
|
ignored = 0;
|
|
|
|
} else if (distance < max_ignorable && xch->ignore) {
|
|
|
|
ignored += xch->chg2;
|
|
|
|
} else if (lxch != xchp &&
|
|
|
|
xch->i1 + ignored - (lxch->i1 + lxch->chg1) > max_common) {
|
|
|
|
break;
|
|
|
|
} else if (!xch->ignore) {
|
|
|
|
lxch = xch;
|
|
|
|
ignored = 0;
|
|
|
|
} else {
|
|
|
|
ignored += xch->chg2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return lxch;
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2022-12-13 11:13:08 +00:00
|
|
|
static long def_ff(const char *rec, long len, char *buf, long sz)
|
2007-07-06 07:45:10 +00:00
|
|
|
{
|
|
|
|
if (len > 0 &&
|
|
|
|
(isalpha((unsigned char)*rec) || /* identifier? */
|
2013-10-16 17:26:39 +00:00
|
|
|
*rec == '_' || /* also identifier? */
|
2007-07-06 07:45:10 +00:00
|
|
|
*rec == '$')) { /* identifiers from VMS and other esoterico */
|
|
|
|
if (len > sz)
|
|
|
|
len = sz;
|
|
|
|
while (0 < len && isspace((unsigned char)rec[len - 1]))
|
|
|
|
len--;
|
|
|
|
memcpy(buf, rec, len);
|
|
|
|
return len;
|
|
|
|
}
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2016-05-28 14:58:47 +00:00
|
|
|
static long match_func_rec(xdfile_t *xdf, xdemitconf_t const *xecfg, long ri,
|
|
|
|
char *buf, long sz)
|
|
|
|
{
|
|
|
|
const char *rec;
|
|
|
|
long len = xdl_get_rec(xdf, ri, &rec);
|
|
|
|
if (!xecfg->find_func)
|
2022-12-13 11:13:08 +00:00
|
|
|
return def_ff(rec, len, buf, sz);
|
2016-05-28 14:58:47 +00:00
|
|
|
return xecfg->find_func(rec, len, buf, sz, xecfg->find_func_priv);
|
|
|
|
}
|
|
|
|
|
2017-11-18 18:04:40 +00:00
|
|
|
static int is_func_rec(xdfile_t *xdf, xdemitconf_t const *xecfg, long ri)
|
|
|
|
{
|
|
|
|
char dummy[1];
|
|
|
|
return match_func_rec(xdf, xecfg, ri, dummy, sizeof(dummy)) >= 0;
|
|
|
|
}
|
|
|
|
|
2011-10-09 11:34:49 +00:00
|
|
|
struct func_line {
|
|
|
|
long len;
|
|
|
|
char buf[80];
|
|
|
|
};
|
|
|
|
|
2011-10-09 11:36:57 +00:00
|
|
|
static long get_func_line(xdfenv_t *xe, xdemitconf_t const *xecfg,
|
2011-10-09 11:34:49 +00:00
|
|
|
struct func_line *func_line, long start, long limit)
|
|
|
|
{
|
2011-10-09 11:36:57 +00:00
|
|
|
long l, size, step = (start > limit) ? -1 : 1;
|
|
|
|
char *buf, dummy[1];
|
2011-10-09 11:34:49 +00:00
|
|
|
|
2011-10-09 11:36:57 +00:00
|
|
|
buf = func_line ? func_line->buf : dummy;
|
|
|
|
size = func_line ? sizeof(func_line->buf) : sizeof(dummy);
|
|
|
|
|
|
|
|
for (l = start; l != limit && 0 <= l && l < xe->xdf1.nrec; l += step) {
|
2016-05-28 14:58:47 +00:00
|
|
|
long len = match_func_rec(&xe->xdf1, xecfg, l, buf, size);
|
2011-10-09 11:34:49 +00:00
|
|
|
if (len >= 0) {
|
2011-10-09 11:36:57 +00:00
|
|
|
if (func_line)
|
|
|
|
func_line->len = len;
|
|
|
|
return l;
|
2011-10-09 11:34:49 +00:00
|
|
|
}
|
|
|
|
}
|
2011-10-09 11:36:57 +00:00
|
|
|
return -1;
|
2011-10-09 11:34:49 +00:00
|
|
|
}
|
|
|
|
|
2016-05-28 15:02:24 +00:00
|
|
|
static int is_empty_rec(xdfile_t *xdf, long ri)
|
|
|
|
{
|
|
|
|
const char *rec;
|
|
|
|
long len = xdl_get_rec(xdf, ri, &rec);
|
|
|
|
|
|
|
|
while (len > 0 && XDL_ISSPACE(*rec)) {
|
|
|
|
rec++;
|
|
|
|
len--;
|
|
|
|
}
|
|
|
|
return !len;
|
|
|
|
}
|
|
|
|
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
int xdl_emit_diff(xdfenv_t *xe, xdchange_t *xscr, xdemitcb_t *ecb,
|
|
|
|
xdemitconf_t const *xecfg) {
|
|
|
|
long s1, s2, e1, e2, lctx;
|
|
|
|
xdchange_t *xch, *xche;
|
2010-09-26 16:26:56 +00:00
|
|
|
long funclineprev = -1;
|
2011-10-09 11:34:49 +00:00
|
|
|
struct func_line func_line = { 0 };
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
|
2009-03-15 21:01:20 +00:00
|
|
|
for (xch = xscr; xch; xch = xche->next) {
|
2019-12-05 16:15:31 +00:00
|
|
|
xdchange_t *xchp = xch;
|
2013-06-19 18:46:07 +00:00
|
|
|
xche = xdl_get_hunk(&xch, xecfg);
|
|
|
|
if (!xch)
|
|
|
|
break;
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
|
2019-12-05 16:15:31 +00:00
|
|
|
pre_context_calculation:
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
s1 = XDL_MAX(xch->i1 - xecfg->ctxlen, 0);
|
|
|
|
s2 = XDL_MAX(xch->i2 - xecfg->ctxlen, 0);
|
|
|
|
|
2011-10-09 11:36:57 +00:00
|
|
|
if (xecfg->flags & XDL_EMIT_FUNCCONTEXT) {
|
2016-05-28 15:00:28 +00:00
|
|
|
long fs1, i1 = xch->i1;
|
|
|
|
|
|
|
|
/* Appended chunk? */
|
|
|
|
if (i1 >= xe->xdf1.nrec) {
|
2016-05-28 15:02:24 +00:00
|
|
|
long i2 = xch->i2;
|
2016-05-28 15:00:28 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We don't need additional context if
|
2017-01-13 16:15:10 +00:00
|
|
|
* a whole function was added.
|
2016-05-28 15:00:28 +00:00
|
|
|
*/
|
2017-01-13 16:15:10 +00:00
|
|
|
while (i2 < xe->xdf2.nrec) {
|
2017-11-18 18:04:40 +00:00
|
|
|
if (is_func_rec(&xe->xdf2, xecfg, i2))
|
2017-01-13 16:15:10 +00:00
|
|
|
goto post_context_calculation;
|
2016-05-28 15:02:24 +00:00
|
|
|
i2++;
|
2017-01-13 16:15:10 +00:00
|
|
|
}
|
2016-05-28 15:00:28 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Otherwise get more context from the
|
|
|
|
* pre-image.
|
|
|
|
*/
|
|
|
|
i1 = xe->xdf1.nrec - 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
fs1 = get_func_line(xe, xecfg, NULL, i1, -1);
|
2017-11-18 18:05:19 +00:00
|
|
|
while (fs1 > 0 && !is_empty_rec(&xe->xdf1, fs1 - 1) &&
|
|
|
|
!is_func_rec(&xe->xdf1, xecfg, fs1 - 1))
|
|
|
|
fs1--;
|
2011-10-09 11:36:57 +00:00
|
|
|
if (fs1 < 0)
|
|
|
|
fs1 = 0;
|
|
|
|
if (fs1 < s1) {
|
2019-07-23 19:27:05 +00:00
|
|
|
s2 = XDL_MAX(s2 - (s1 - fs1), 0);
|
2011-10-09 11:36:57 +00:00
|
|
|
s1 = fs1;
|
2019-12-05 16:15:31 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Did we extend context upwards into an
|
|
|
|
* ignored change?
|
|
|
|
*/
|
|
|
|
while (xchp != xch &&
|
|
|
|
xchp->i1 + xchp->chg1 <= s1 &&
|
|
|
|
xchp->i2 + xchp->chg2 <= s2)
|
|
|
|
xchp = xchp->next;
|
|
|
|
|
|
|
|
/* If so, show it after all. */
|
|
|
|
if (xchp != xch) {
|
|
|
|
xch = xchp;
|
|
|
|
goto pre_context_calculation;
|
|
|
|
}
|
2011-10-09 11:36:57 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-05-28 15:00:28 +00:00
|
|
|
post_context_calculation:
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
lctx = xecfg->ctxlen;
|
|
|
|
lctx = XDL_MIN(lctx, xe->xdf1.nrec - (xche->i1 + xche->chg1));
|
|
|
|
lctx = XDL_MIN(lctx, xe->xdf2.nrec - (xche->i2 + xche->chg2));
|
|
|
|
|
|
|
|
e1 = xche->i1 + xche->chg1 + lctx;
|
|
|
|
e2 = xche->i2 + xche->chg2 + lctx;
|
|
|
|
|
2011-10-09 11:36:57 +00:00
|
|
|
if (xecfg->flags & XDL_EMIT_FUNCCONTEXT) {
|
|
|
|
long fe1 = get_func_line(xe, xecfg, NULL,
|
|
|
|
xche->i1 + xche->chg1,
|
|
|
|
xe->xdf1.nrec);
|
2016-05-28 15:03:16 +00:00
|
|
|
while (fe1 > 0 && is_empty_rec(&xe->xdf1, fe1 - 1))
|
|
|
|
fe1--;
|
2011-10-09 11:36:57 +00:00
|
|
|
if (fe1 < 0)
|
|
|
|
fe1 = xe->xdf1.nrec;
|
|
|
|
if (fe1 > e1) {
|
2019-07-23 19:27:05 +00:00
|
|
|
e2 = XDL_MIN(e2 + (fe1 - e1), xe->xdf2.nrec);
|
2011-10-09 11:36:57 +00:00
|
|
|
e1 = fe1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Overlap with next change? Then include it
|
|
|
|
* in the current hunk and start over to find
|
|
|
|
* its new end.
|
|
|
|
*/
|
|
|
|
if (xche->next) {
|
2016-06-09 21:54:48 +00:00
|
|
|
long l = XDL_MIN(xche->next->i1,
|
|
|
|
xe->xdf1.nrec - 1);
|
2016-09-14 16:05:27 +00:00
|
|
|
if (l - xecfg->ctxlen <= e1 ||
|
2011-10-09 11:36:57 +00:00
|
|
|
get_func_line(xe, xecfg, NULL, l, e1) < 0) {
|
|
|
|
xche = xche->next;
|
2016-05-28 15:00:28 +00:00
|
|
|
goto post_context_calculation;
|
2011-10-09 11:36:57 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
/*
|
|
|
|
* Emit current hunk header.
|
|
|
|
*/
|
2006-03-28 02:23:31 +00:00
|
|
|
|
|
|
|
if (xecfg->flags & XDL_EMIT_FUNCNAMES) {
|
2011-10-09 11:34:49 +00:00
|
|
|
get_func_line(xe, xecfg, &func_line,
|
|
|
|
s1 - 1, funclineprev);
|
2010-09-26 16:26:56 +00:00
|
|
|
funclineprev = s1 - 1;
|
2006-03-28 02:23:31 +00:00
|
|
|
}
|
2021-04-12 17:15:29 +00:00
|
|
|
if (!(xecfg->flags & XDL_EMIT_NO_HUNK_HDR) &&
|
|
|
|
xdl_emit_hunk_hdr(s1 + 1, e1 - s1, s2 + 1, e2 - s2,
|
2011-10-09 11:34:49 +00:00
|
|
|
func_line.buf, func_line.len, ecb) < 0)
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Emit pre-context.
|
|
|
|
*/
|
2012-01-06 17:13:00 +00:00
|
|
|
for (; s2 < xch->i2; s2++)
|
|
|
|
if (xdl_emit_record(&xe->xdf2, s2, " ", ecb) < 0)
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
for (s1 = xch->i1, s2 = xch->i2;; xch = xch->next) {
|
|
|
|
/*
|
|
|
|
* Merge previous with current change atom.
|
|
|
|
*/
|
|
|
|
for (; s1 < xch->i1 && s2 < xch->i2; s1++, s2++)
|
2012-01-06 17:13:00 +00:00
|
|
|
if (xdl_emit_record(&xe->xdf2, s2, " ", ecb) < 0)
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Removes lines from the first file.
|
|
|
|
*/
|
|
|
|
for (s1 = xch->i1; s1 < xch->i1 + xch->chg1; s1++)
|
|
|
|
if (xdl_emit_record(&xe->xdf1, s1, "-", ecb) < 0)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Adds lines from the second file.
|
|
|
|
*/
|
|
|
|
for (s2 = xch->i2; s2 < xch->i2 + xch->chg2; s2++)
|
|
|
|
if (xdl_emit_record(&xe->xdf2, s2, "+", ecb) < 0)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
if (xch == xche)
|
|
|
|
break;
|
|
|
|
s1 = xch->i1 + xch->chg1;
|
|
|
|
s2 = xch->i2 + xch->chg2;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Emit post-context.
|
|
|
|
*/
|
2012-01-06 17:13:00 +00:00
|
|
|
for (s2 = xche->i2 + xche->chg2; s2 < e2; s2++)
|
|
|
|
if (xdl_emit_record(&xe->xdf2, s2, " ", ecb) < 0)
|
Use a *real* built-in diff generator
This uses a simplified libxdiff setup to generate unified diffs _without_
doing fork/execve of GNU "diff".
This has several huge advantages, for example:
Before:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m24.818s
user 0m13.332s
sys 0m8.664s
After:
[torvalds@g5 linux]$ time git diff v2.6.16.. > /dev/null
real 0m4.563s
user 0m2.944s
sys 0m1.580s
and the fact that this should be a lot more portable (ie we can ignore all
the issues with doing fork/execve under Windows).
Perhaps even more importantly, this allows us to do diffs without actually
ever writing out the git file contents to a temporary file (and without
any of the shell quoting issues on filenames etc etc).
NOTE! THIS PATCH DOES NOT DO THAT OPTIMIZATION YET! I was lazy, and the
current "diff-core" code actually will always write the temp-files,
because it used to be something that you simply had to do. So this current
one actually writes a temp-file like before, and then reads it into memory
again just to do the diff. Stupid.
But if this basic infrastructure is accepted, we can start switching over
diff-core to not write temp-files, which should speed things up even
further, especially when doing big tree-to-tree diffs.
Now, in the interest of full disclosure, I should also point out a few
downsides:
- the libxdiff algorithm is different, and I bet GNU diff has gotten a
lot more testing. And the thing is, generating a diff is not an exact
science - you can get two different diffs (and you will), and they can
both be perfectly valid. So it's not possible to "validate" the
libxdiff output by just comparing it against GNU diff.
- GNU diff does some nice eye-candy, like trying to figure out what the
last function was, and adding that information to the "@@ .." line.
libxdiff doesn't do that.
- The libxdiff thing has some known deficiencies. In particular, it gets
the "\No newline at end of file" case wrong. So this is currently for
the experimental branch only. I hope Davide will help fix it.
That said, I think the huge performance advantage, and the fact that it
integrates better is definitely worth it. But it should go into a
development branch at least due to the missing newline issue.
Technical note: this is based on libxdiff-0.17, but I did some surgery to
get rid of the extraneous fat - stuff that git doesn't need, and seriously
cutting down on mmfile_t, which had much more capabilities than the diff
algorithm either needed or used. In this version, "mmfile_t" is just a
trivial <pointer,length> tuple.
That said, I tried to keep the differences to simple removals, so that you
can do a diff between this and the libxdiff origin, and you'll basically
see just things getting deleted. Even the mmfile_t simplifications are
left in a state where the diffs should be readable.
Apologies to Davide, whom I'd love to get feedback on this all from (I
wrote my own "fill_mmfile()" for the new simpler mmfile_t format: the old
complex format had a helper function for that, but I did my surgery with
the goal in mind that eventually we _should_ just do
mmfile_t mf;
buf = read_sha1_file(sha1, type, &size);
mf->ptr = buf;
mf->size = size;
.. use "mf" directly ..
which was really a nightmare with the old "helpful" mmfile_t, and really
is that easy with the new cut-down interfaces).
[ Btw, as any hawk-eye can see from the diff, this was actually generated
with itself, so it is "self-hosting". That's about all the testing it
has gotten, along with the above kernel diff, which eye-balls correctly,
but shows the newline issue when you double-check it with "git-apply" ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-03-25 04:13:22 +00:00
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|