git/lib/blame.tcl

874 lines
22 KiB
Tcl
Raw Normal View History

# git-gui blame viewer
# Copyright (C) 2006, 2007 Shawn Pearce
class blame {
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
image create photo ::blame::img_back_arrow -data {R0lGODlhGAAYAIUAAPwCBEzKXFTSZIz+nGzmhGzqfGTidIT+nEzGXHTqhGzmfGzifFzadETCVES+VARWDFzWbHzyjAReDGTadFTOZDSyRDyyTCymPARaFGTedFzSbDy2TCyqRCyqPARaDAyCHES6VDy6VCyiPAR6HCSeNByWLARyFARiDARqFGTifARiFAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACH5BAEAAAAALAAAAAAYABgAAAajQIBwSCwaj8ikcsk0BppJwRPqHEypQwHBis0WDAdEFyBIKBaMAKLBdjQeSkFBYTBAIvgEoS6JmhUTEwIUDQ4VFhcMGEhyCgoZExoUaxsWHB0THkgfAXUGAhoBDSAVFR0XBnCbDRmgog0hpSIiDJpJIyEQhBUcJCIlwA22SSYVogknEg8eD82qSigdDSknY0IqJQXPYxIl1dZCGNvWw+Dm510GQQAh/mhDcmVhdGVkIGJ5IEJNUFRvR0lGIFBybyB2ZXJzaW9uIDIuNQ0KqSBEZXZlbENvciAxOTk3LDE5OTguIEFsbCByaWdodHMgcmVzZXJ2ZWQuDQpodHRwOi8vd3d3LmRldmVsY29yLmNvbQA7}
field commit ; # input commit to blame
field path ; # input filename to view in $commit
field history {}; # viewer history: {commit path}
field w ; # top window in this viewer
field w_back ; # our back button
field w_path ; # label showing the current file path
field w_line ; # text column: all line numbers
field w_cgrp ; # text column: abbreviated commit SHA-1s
field w_load ; # text column: loaded indicator
field w_file ; # text column: actual file data
field w_cmit ; # pane showing commit message
field status ; # text variable bound to status bar
field old_height ; # last known height of $w.file_pane
field current_fd {} ; # background process running
field highlight_line -1 ; # current line selected
field highlight_commit {} ; # sha1 of commit selected
field total_lines 0 ; # total length of file
field blame_lines 0 ; # number of lines computed
field commit_count 0 ; # number of commits in $commit_list
field commit_list {} ; # list of commit sha1 in receipt order
field order ; # array commit -> receipt order
field header ; # array commit,key -> header field
field line_commit ; # array line -> sha1 commit
field line_file ; # array line -> file name
field r_commit ; # commit currently being parsed
field r_orig_line ; # original line number
field r_final_line ; # final line number
field r_line_count ; # lines in this region
field tooltip_wm {} ; # Current tooltip toplevel, if open
field tooltip_timer {} ; # Current timer event for our tooltip
field tooltip_commit {} ; # Commit in tooltip
field tooltip_text {} ; # Text in current tooltip
variable active_color #98e1a0
variable group_colors {
#cbcbcb
#e1e1e1
}
constructor new {i_commit i_path} {
variable active_color
global cursor_ptr
set commit $i_commit
set path $i_path
make_toplevel top w
wm title $top "[appname] ([reponame]): File Viewer"
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
frame $w.header -background orange
label $w.header.commit_l \
-text {Commit:} \
-background orange \
-anchor w \
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
-justify left
set w_back $w.header.commit_b
button $w_back \
-command [cb _history_menu] \
-image ::blame::img_back_arrow \
-borderwidth 0 \
-relief flat \
-state disabled \
-background orange \
-activebackground orange
label $w.header.commit \
-textvariable @commit \
-background orange \
-anchor w \
-justify left
label $w.header.path_l \
-text {File:} \
-background orange \
-anchor w \
-justify left
set w_path $w.header.path
label $w_path \
-background orange \
-anchor w \
-justify left
pack $w.header.commit_l -side left
pack $w_back -side left
pack $w.header.commit -side left
pack $w_path -fill x -side right
pack $w.header.path_l -side right
panedwindow $w.file_pane -orient vertical
frame $w.file_pane.out
frame $w.file_pane.cm
$w.file_pane add $w.file_pane.out \
-sticky nsew \
-minsize 100 \
-height 100 \
-width 100
$w.file_pane add $w.file_pane.cm \
-sticky nsew \
-minsize 25 \
-height 25 \
-width 100
set w_load $w.file_pane.out.loaded_t
text $w_load \
-background white -borderwidth 0 \
-state disabled \
-wrap none \
-height 40 \
-width 1 \
-font font_diff
$w_load tag conf annotated -background grey
set w_line $w.file_pane.out.linenumber_t
text $w_line \
-background white -borderwidth 0 \
-state disabled \
-wrap none \
-height 40 \
-width 5 \
-font font_diff
$w_line tag conf linenumber -justify right
set w_cgrp $w.file_pane.out.commit_t
text $w_cgrp \
-background white -borderwidth 0 \
-state disabled \
-wrap none \
-height 40 \
-width 4 \
-font font_diff
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
$w_cgrp tag conf curr_commit
$w_cgrp tag conf prior_commit \
-foreground blue \
-underline 1
$w_cgrp tag bind prior_commit \
<Button-1> \
"[cb _load_commit @%x,%y];break"
set w_file $w.file_pane.out.file_t
text $w_file \
-background white -borderwidth 0 \
-state disabled \
-wrap none \
-height 40 \
-width 80 \
-xscrollcommand [list $w.file_pane.out.sbx set] \
-font font_diff
scrollbar $w.file_pane.out.sbx \
-orient h \
-command [list $w_file xview]
scrollbar $w.file_pane.out.sby \
-orient v \
-command [list scrollbar2many [list \
$w_load \
$w_line \
$w_cgrp \
$w_file \
] yview]
grid \
$w_cgrp \
$w_line \
$w_load \
$w_file \
$w.file_pane.out.sby \
-sticky nsew
grid conf $w.file_pane.out.sbx -column 3 -sticky we
grid columnconfigure $w.file_pane.out 3 -weight 1
grid rowconfigure $w.file_pane.out 0 -weight 1
set w_cmit $w.file_pane.cm.t
text $w_cmit \
-background white -borderwidth 0 \
-state disabled \
-wrap none \
-height 10 \
-width 80 \
-xscrollcommand [list $w.file_pane.cm.sbx set] \
-yscrollcommand [list $w.file_pane.cm.sby set] \
-font font_diff
$w_cmit tag conf header_key \
-tabs {3c} \
-background $active_color \
-font font_uibold
$w_cmit tag conf header_val \
-background $active_color \
-font font_ui
$w_cmit tag raise sel
scrollbar $w.file_pane.cm.sbx \
-orient h \
-command [list $w_cmit xview]
scrollbar $w.file_pane.cm.sby \
-orient v \
-command [list $w_cmit yview]
pack $w.file_pane.cm.sby -side right -fill y
pack $w.file_pane.cm.sbx -side bottom -fill x
pack $w_cmit -expand 1 -fill both
frame $w.status \
-borderwidth 1 \
-relief sunken
label $w.status.l \
-textvariable @status \
-anchor w \
-justify left
pack $w.status.l -side left
menu $w.ctxm -tearoff 0
$w.ctxm add command \
-label "Copy Commit" \
-command [cb _copycommit]
foreach i [list \
$w_cgrp \
$w_load \
$w_line \
$w_file] {
$i conf -cursor $cursor_ptr
$i conf -yscrollcommand \
[list many2scrollbar [list \
$w_cgrp \
$w_load \
$w_line \
$w_file \
] yview $w.file_pane.out.sby]
bind $i <Button-1> "
[cb _hide_tooltip]
[cb _click $i @%x,%y]
focus $i
"
bind $i <Any-Motion> [cb _show_tooltip $i @%x,%y]
bind $i <Any-Enter> [cb _hide_tooltip]
bind $i <Any-Leave> [cb _hide_tooltip]
bind_button3 $i "
[cb _hide_tooltip]
set cursorX %x
set cursorY %y
set cursorW %W
tk_popup $w.ctxm %X %Y
"
}
foreach i [list \
$w_cgrp \
$w_load \
$w_line \
$w_file \
$w_cmit] {
bind $i <Key-Up> {catch {%W yview scroll -1 units};break}
bind $i <Key-Down> {catch {%W yview scroll 1 units};break}
bind $i <Key-Left> {catch {%W xview scroll -1 units};break}
bind $i <Key-Right> {catch {%W xview scroll 1 units};break}
bind $i <Key-k> {catch {%W yview scroll -1 units};break}
bind $i <Key-j> {catch {%W yview scroll 1 units};break}
bind $i <Key-h> {catch {%W xview scroll -1 units};break}
bind $i <Key-l> {catch {%W xview scroll 1 units};break}
bind $i <Control-Key-b> {catch {%W yview scroll -1 pages};break}
bind $i <Control-Key-f> {catch {%W yview scroll 1 pages};break}
}
bind $w_cmit <Button-1> [list focus $w_cmit]
bind $top <Visibility> [list focus $top]
bind $w_file <Destroy> [list delete_this $this]
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
grid configure $w.header -sticky ew
grid configure $w.file_pane -sticky nsew
grid configure $w.status -sticky ew
grid columnconfigure $top 0 -weight 1
grid rowconfigure $top 0 -weight 0
grid rowconfigure $top 1 -weight 1
grid rowconfigure $top 2 -weight 0
set req_w [winfo reqwidth $top]
set req_h [winfo reqheight $top]
if {$req_w < 600} {set req_w 600}
if {$req_h < 400} {set req_h 400}
set g "${req_w}x${req_h}"
wm geometry $top $g
update
set old_height [winfo height $w.file_pane]
$w.file_pane sash place 0 \
[lindex [$w.file_pane sash coord 0] 0] \
[expr {int($old_height * 0.70)}]
bind $w.file_pane <Configure> \
"if {{$w.file_pane} eq {%W}} {[cb _resize %h]}"
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
_load $this
}
method _load {} {
_hide_tooltip $this
if {$total_lines != 0 || $current_fd ne {}} {
if {$current_fd ne {}} {
catch {close $current_fd}
set current_fd {}
}
set highlight_line -1
set highlight_commit {}
set total_lines 0
set blame_lines 0
set commit_count 0
set commit_list {}
array unset order
array unset line_commit
array unset line_file
$w_load conf -state normal
$w_cgrp conf -state normal
$w_line conf -state normal
$w_file conf -state normal
$w_load delete 0.0 end
$w_cgrp delete 0.0 end
$w_line delete 0.0 end
$w_file delete 0.0 end
$w_load conf -state disabled
$w_cgrp conf -state disabled
$w_line conf -state disabled
$w_file conf -state disabled
}
if {[winfo exists $w.status.c]} {
$w.status.c coords bar 0 0 0 20
} else {
canvas $w.status.c \
-width 100 \
-height [expr {int([winfo reqheight $w.status.l] * 0.6)}] \
-borderwidth 1 \
-relief groove \
-highlightt 0
$w.status.c create rectangle 0 0 0 20 -tags bar -fill navy
pack $w.status.c -side right
}
if {$history eq {}} {
$w_back conf -state disabled
} else {
$w_back conf -state normal
}
lappend history [list $commit $path]
set status "Loading $commit:[escape_path $path]..."
$w_path conf -text [escape_path $path]
if {$commit eq {}} {
set fd [open $path r]
} else {
set cmd [list git cat-file blob "$commit:$path"]
set fd [open "| $cmd" r]
}
fconfigure $fd -blocking 0 -translation lf -encoding binary
fileevent $fd readable [cb _read_file $fd]
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
set current_fd $fd
}
method _history_menu {} {
set m $w.backmenu
if {[winfo exists $m]} {
$m delete 0 end
} else {
menu $m -tearoff 0
}
for {set i [expr {[llength $history] - 2}]
} {$i >= 0} {incr i -1} {
set e [lindex $history $i]
set c [lindex $e 0]
set f [lindex $e 1]
if {[regexp {^[0-9a-f]{40}$} $c]} {
set t [string range $c 0 8]...
} else {
set t $c
}
if {![catch {set summary $header($c,summary)}]} {
append t " $summary"
}
$m add command -label $t -command [cb _goback $i $c $f]
}
set X [winfo rootx $w_back]
set Y [expr {[winfo rooty $w_back] + [winfo height $w_back]}]
tk_popup $m $X $Y
}
method _goback {i c f} {
set history [lrange $history 0 [expr {$i - 1}]]
set commit $c
set path $f
_load $this
}
method _read_file {fd} {
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
if {$fd ne $current_fd} {
catch {close $fd}
return
}
$w_load conf -state normal
$w_cgrp conf -state normal
$w_line conf -state normal
$w_file conf -state normal
while {[gets $fd line] >= 0} {
regsub "\r\$" $line {} line
incr total_lines
if {$total_lines > 1} {
$w_load insert end "\n"
$w_cgrp insert end "\n"
$w_line insert end "\n"
$w_file insert end "\n"
}
$w_line insert end "$total_lines" linenumber
$w_file insert end "$line"
}
$w_load conf -state disabled
$w_cgrp conf -state disabled
$w_line conf -state disabled
$w_file conf -state disabled
if {[eof $fd]} {
close $fd
_status $this
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
set cmd {nice git blame -M -C --incremental}
if {$commit eq {}} {
lappend cmd --contents $path
} else {
lappend cmd $commit
}
lappend cmd -- $path
set fd [open "| $cmd" r]
fconfigure $fd -blocking 0 -translation lf -encoding binary
fileevent $fd readable [cb _read_blame $fd]
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
set current_fd $fd
}
} ifdeleted { catch {close $fd} }
method _read_blame {fd} {
variable group_colors
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
if {$fd ne $current_fd} {
catch {close $fd}
return
}
$w_cgrp conf -state normal
while {[gets $fd line] >= 0} {
if {[regexp {^([a-z0-9]{40}) (\d+) (\d+) (\d+)$} $line line \
cmit original_line final_line line_count]} {
set r_commit $cmit
set r_orig_line $original_line
set r_final_line $final_line
set r_line_count $line_count
if {[catch {set g $order($cmit)}]} {
set bg [lindex $group_colors 0]
set group_colors [lrange $group_colors 1 end]
lappend group_colors $bg
$w_cgrp tag conf g$cmit -background $bg
$w_line tag conf g$cmit -background $bg
$w_file tag conf g$cmit -background $bg
set order($cmit) $commit_count
incr commit_count
lappend commit_list $cmit
}
} elseif {[string match {filename *} $line]} {
set file [string range $line 9 end]
set n $r_line_count
set lno $r_final_line
set cmit $r_commit
if {[regexp {^0{40}$} $cmit]} {
set commit_abbr work
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
set commit_type curr_commit
} elseif {$cmit eq $commit} {
set commit_abbr this
set commit_type curr_commit
} else {
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
set commit_type prior_commit
set commit_abbr [string range $cmit 0 4]
}
set author_abbr {}
set a_name {}
catch {set a_name $header($cmit,author)}
while {$a_name ne {}} {
if {![regexp {^([[:upper:]])} $a_name _a]} break
append author_abbr $_a
unset _a
if {![regsub \
{^[[:upper:]][^\s]*\s+} \
$a_name {} a_name ]} break
}
if {$author_abbr eq {}} {
set author_abbr { |}
} else {
set author_abbr [string range $author_abbr 0 3]
while {[string length $author_abbr] < 4} {
set author_abbr " $author_abbr"
}
}
unset a_name
set first_lno $lno
while {
![catch {set ncmit $line_commit([expr {$first_lno - 1}])}]
&& ![catch {set nfile $line_file([expr {$first_lno - 1}])}]
&& $ncmit eq $cmit
&& $nfile eq $file
} {
incr first_lno -1
}
while {$n > 0} {
set lno_e "$lno.0 lineend + 1c"
if {[catch {set g g$line_commit($lno)}]} {
$w_load tag add annotated $lno.0 $lno_e
} else {
$w_cgrp tag remove g$g $lno.0 $lno_e
$w_line tag remove g$g $lno.0 $lno_e
$w_file tag remove g$g $lno.0 $lno_e
$w_cgrp tag remove a$g $lno.0 $lno_e
$w_line tag remove a$g $lno.0 $lno_e
$w_file tag remove a$g $lno.0 $lno_e
}
set line_commit($lno) $cmit
set line_file($lno) $file
$w_cgrp delete $lno.0 "$lno.0 lineend"
if {$lno == $first_lno} {
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
$w_cgrp insert $lno.0 $commit_abbr $commit_type
} elseif {$lno == [expr {$first_lno + 1}]} {
$w_cgrp insert $lno.0 $author_abbr
} else {
$w_cgrp insert $lno.0 { |}
}
$w_cgrp tag add g$cmit $lno.0 $lno_e
$w_line tag add g$cmit $lno.0 $lno_e
$w_file tag add g$cmit $lno.0 $lno_e
$w_cgrp tag add a$cmit $lno.0 $lno_e
$w_line tag add a$cmit $lno.0 $lno_e
$w_file tag add a$cmit $lno.0 $lno_e
if {$highlight_line == -1} {
if {[lindex [$w_file yview] 0] == 0} {
$w_file see $lno.0
_showcommit $this $lno
}
} elseif {$highlight_line == $lno} {
_showcommit $this $lno
}
incr n -1
incr lno
incr blame_lines
}
while {
![catch {set ncmit $line_commit($lno)}]
&& ![catch {set nfile $line_file($lno)}]
&& $ncmit eq $cmit
&& $nfile eq $file
} {
$w_cgrp delete $lno.0 "$lno.0 lineend"
if {$lno == $first_lno} {
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
$w_cgrp insert $lno.0 $commit_abbr $commit_type
} elseif {$lno == [expr {$first_lno + 1}]} {
$w_cgrp insert $lno.0 $author_abbr
} else {
$w_cgrp insert $lno.0 { |}
}
incr lno
}
} elseif {[regexp {^([a-z-]+) (.*)$} $line line key data]} {
set header($r_commit,$key) $data
}
}
$w_cgrp conf -state disabled
if {[eof $fd]} {
close $fd
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
set current_fd {}
set status {Annotation complete.}
destroy $w.status.c
} else {
_status $this
}
} ifdeleted { catch {close $fd} }
method _status {} {
set have $blame_lines
set total $total_lines
set pdone 0
if {$total} {set pdone [expr {100 * $have / $total}]}
set status [format \
"Loading annotations... %i of %i lines annotated (%2i%%)" \
$have $total $pdone]
$w.status.c coords bar 0 0 $pdone 20
}
method _click {cur_w pos} {
set lno [lindex [split [$cur_w index $pos] .] 0]
if {$lno eq {}} return
_showcommit $this $lno
}
git-gui: Allow digging through history in blame viewer gitweb has long had a feature where the user can click on any commit the blame display and go visit that commit's information page. From the user could go get the blame display for the file they are tracking, and try to digg through the history of any part of the code they are interested in seeing. We now offer somewhat similiar functionality in git-gui. The 4 digit commit abreviation in the first column of our blame view is now offered as a hyperlink if the commit isn't the one we are now viewing the blame output for (as there is no point in linking back to yourself). Clicking on that link will stop the current blame engine (if still running), push the new target commit onto the history stack, and restart the blame viewer at that commit, using the "original file name" as supplied by git-blame for that chunk of the output. Users can navigate back to a version they had been viewing before by way of a back button, which offers the prior commits in a popup menu displayed right below the back button. I'm always showing the menu here as the cost of switching between views is very high; you don't want to jump to a commit you are not interested in looking at again. During switches we throw away all data except the cached commit data, as that is relatively small compared to most source files and their annotation marks. Unfortunately throwing this per-file data away in Tcl seems to take some time; I probably should move the line indexed arrays to proper lists and use [lindex] rather than the array lookup (usually lists are faster). We now start the git-blame process using "nice", so that its priority will drop hopefully below our own. If I don't do this the blame engine gets a lot of CPU under Windows 2000 and the git-gui user interface is almost non-responsive, even though Tcl is just sitting there waiting for events. Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
2007-06-02 01:47:12 +00:00
method _load_commit {pos} {
set lno [lindex [split [$w_cgrp index $pos] .] 0]
if {[catch {set cmit $line_commit($lno)}]} return
if {[catch {set file $line_file($lno) }]} return
set commit $cmit
set path $file
_load $this
}
method _showcommit {lno} {
global repo_config
variable active_color
if {$highlight_commit ne {}} {
set cmit $highlight_commit
$w_cgrp tag conf a$cmit -background {}
$w_line tag conf a$cmit -background {}
$w_file tag conf a$cmit -background {}
}
$w_cmit conf -state normal
$w_cmit delete 0.0 end
if {[catch {set cmit $line_commit($lno)}]} {
set cmit {}
$w_cmit insert end "Loading annotation..."
} else {
$w_cgrp tag conf a$cmit -background $active_color
$w_line tag conf a$cmit -background $active_color
$w_file tag conf a$cmit -background $active_color
set author_name {}
set author_email {}
set author_time {}
catch {set author_name $header($cmit,author)}
catch {set author_email $header($cmit,author-mail)}
catch {set author_time [clock format \
$header($cmit,author-time) \
-format {%Y-%m-%d %H:%M:%S}
]}
set committer_name {}
set committer_email {}
set committer_time {}
catch {set committer_name $header($cmit,committer)}
catch {set committer_email $header($cmit,committer-mail)}
catch {set committer_time [clock format \
$header($cmit,committer-time) \
-format {%Y-%m-%d %H:%M:%S}
]}
if {[catch {set msg $header($cmit,message)}]} {
set msg {}
catch {
set fd [open "| git cat-file commit $cmit" r]
fconfigure $fd -encoding binary -translation lf
if {[catch {set enc $repo_config(i18n.commitencoding)}]} {
set enc utf-8
}
while {[gets $fd line] > 0} {
if {[string match {encoding *} $line]} {
set enc [string tolower [string range $line 9 end]]
}
}
set msg [encoding convertfrom $enc [read $fd]]
set msg [string trim $msg]
close $fd
set author_name [encoding convertfrom $enc $author_name]
set committer_name [encoding convertfrom $enc $committer_name]
set header($cmit,author) $author_name
set header($cmit,committer) $committer_name
}
set header($cmit,message) $msg
}
$w_cmit insert end "commit $cmit\n" header_key
$w_cmit insert end "Author:\t" header_key
$w_cmit insert end "$author_name $author_email" header_val
$w_cmit insert end "$author_time\n" header_val
$w_cmit insert end "Committer:\t" header_key
$w_cmit insert end "$committer_name $committer_email" header_val
$w_cmit insert end "$committer_time\n" header_val
if {$line_file($lno) ne $path} {
$w_cmit insert end "Original File:\t" header_key
$w_cmit insert end "[escape_path $line_file($lno)]\n" header_val
}
$w_cmit insert end "\n$msg"
}
$w_cmit conf -state disabled
set highlight_line $lno
set highlight_commit $cmit
if {$highlight_commit eq $tooltip_commit} {
_hide_tooltip $this
}
}
method _copycommit {} {
set pos @$::cursorX,$::cursorY
set lno [lindex [split [$::cursorW index $pos] .] 0]
if {![catch {set commit $line_commit($lno)}]} {
clipboard clear
clipboard append \
-format STRING \
-type STRING \
-- $commit
}
}
method _show_tooltip {cur_w pos} {
set lno [lindex [split [$cur_w index $pos] .] 0]
if {[catch {set cmit $line_commit($lno)}]} {
_hide_tooltip $this
return
}
if {$cmit eq $highlight_commit} {
_hide_tooltip $this
return
}
if {$cmit eq $tooltip_commit} {
_position_tooltip $this
} elseif {$tooltip_wm ne {}} {
_open_tooltip $this $cur_w
} elseif {$tooltip_timer eq {}} {
set tooltip_timer [after 1000 [cb _open_tooltip $cur_w]]
}
}
method _open_tooltip {cur_w} {
set tooltip_timer {}
set pos_x [winfo pointerx $cur_w]
set pos_y [winfo pointery $cur_w]
if {[winfo containing $pos_x $pos_y] ne $cur_w} {
_hide_tooltip $this
return
}
set pos @[join [list \
[expr {$pos_x - [winfo rootx $cur_w]}] \
[expr {$pos_y - [winfo rooty $cur_w]}]] ,]
set lno [lindex [split [$cur_w index $pos] .] 0]
set cmit $line_commit($lno)
set author_name {}
set author_email {}
set author_time {}
catch {set author_name $header($cmit,author)}
catch {set author_email $header($cmit,author-mail)}
catch {set author_time [clock format \
$header($cmit,author-time) \
-format {%Y-%m-%d %H:%M:%S}
]}
set committer_name {}
set committer_email {}
set committer_time {}
catch {set committer_name $header($cmit,committer)}
catch {set committer_email $header($cmit,committer-mail)}
catch {set committer_time [clock format \
$header($cmit,committer-time) \
-format {%Y-%m-%d %H:%M:%S}
]}
set summary {}
catch {set summary $header($cmit,summary)}
set tooltip_commit $cmit
set tooltip_text "commit $cmit
$author_name $author_email $author_time
$summary"
set file $line_file($lno)
if {$file ne $path} {
append tooltip_text "
Original File: $file"
}
if {$tooltip_wm ne "$cur_w.tooltip"} {
_hide_tooltip $this
set tooltip_wm [toplevel $cur_w.tooltip -borderwidth 1]
wm overrideredirect $tooltip_wm 1
wm transient $tooltip_wm [winfo toplevel $cur_w]
pack [label $tooltip_wm.label \
-background lightyellow \
-foreground black \
-textvariable @tooltip_text \
-justify left]
}
_position_tooltip $this
}
method _position_tooltip {} {
set req_w [winfo reqwidth $tooltip_wm.label]
set req_h [winfo reqheight $tooltip_wm.label]
set pos_x [expr {[winfo pointerx .] + 5}]
set pos_y [expr {[winfo pointery .] + 10}]
set g "${req_w}x${req_h}"
if {$pos_x >= 0} {append g +}
append g $pos_x
if {$pos_y >= 0} {append g +}
append g $pos_y
wm geometry $tooltip_wm $g
raise $tooltip_wm
}
method _hide_tooltip {} {
if {$tooltip_wm ne {}} {
destroy $tooltip_wm
set tooltip_wm {}
set tooltip_commit {}
}
if {$tooltip_timer ne {}} {
after cancel $tooltip_timer
set tooltip_timer {}
}
}
method _resize {new_height} {
set diff [expr {$new_height - $old_height}]
if {$diff == 0} return
set my [expr {[winfo height $w.file_pane] - 25}]
set o [$w.file_pane sash coord 0]
set ox [lindex $o 0]
set oy [expr {[lindex $o 1] + $diff}]
if {$oy < 0} {set oy 0}
if {$oy > $my} {set oy $my}
$w.file_pane sash place 0 $ox $oy
set old_height $new_height
}
}