Assume that we want to commit these states:
Old state == HEAD Intermediate state New state
--------------------------------------------------------
context before context before context before
old 1 new 1 new 1
old 2 old 2 new 2
context after context after context after
that is, want to commit two changes in this order:
1. transform "old 1" into "new 1"
2. transform "old 2" into "new 2"
[This discussion and this patch is about this very case and one other case
as outlined below; any other intermediate states that one could imagine are
not affected by this patch.]
Now assume further, that we have not staged and commited anything, but we
have already changed the working file to the new state. Then we will see
this hunk in the "Unstaged Changes":
@@ -1,4 +1,4 @@
context before
-old 1
-old 2
+new 1
+new 2
context after
The obvious way to stage the intermediate state is to apply "Stage This
Line" to "-old 1" and "+new 1". Unfortunately, this resulted in this
intermediate state:
context before
old 2
new 1
context after
which is not what we wanted. In fact, it was impossible to stage the
intermediate state using "Stage Line". The crux was that if a "+" line was
staged, then the "-" lines were converted to context lines and arranged
*before* the "+" line in the forged hunk that we fed to 'git apply'.
With this patch we now treat "+" lines that are staged differently. In
particular, the "-" lines before the "+" block are moved *after* the
staged "+" line. Now it is possible to get the correct intermediate state
by staging "-old 1" and "+new 1". Problem solved.
But there is a catch.
Noticing that we didn't get the right intermediate state by staging
"-old 1" and "+new 1", we could have had the idea to stage the complete
hunk and to *unstage* "-old 2" and "+new 2". But... the result is the same.
The reason is that there is the exact symmetric problem with unstaging the
last "-" and "+" line that are in adjacent blocks of "-" and "+" lines.
This patch does *not* change the way in which "-" lines are *unstaged*.
Why? Because if we did (i.e. move "+" lines before the "-" line after
converting them to context lines), then it would be impossible to stage
this intermediate state:
context before
old 1
new 2
context after
that is, it would be impossible to stage the two independet changes in the
opposite order.
Let's look at this case a bit further: The obvious way to get this
intermediate state would be to apply "Stage This Line" to "-old 2" and
"+new 2". Before this patch, this worked as expected. With this patch, it
does not work as expected, but it can still be achieved by first staging
the entire hunk, then *unstaging* "-old 1" and "+new 1".
In summary, this patch makes a common case possible, at the expense that
a less common case is made more complicated for the user.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
To "Stage/Unstage Line" we construct a patch that contains exactly one
change (either addition or removal); the hunk header was forged by counting
the old side and adjusting the count by +/-1 for the new side. But when we
counted the context we never counted the changed line itself. If the hunk
had only one removal line and one line of context, like this:
@@ -1,3 +1,2 @@
context 1
-removal
context 2
We had constructed this patch:
@@ -1,2 +1,1 @@
context 1
-removal
context 2
which does not apply because git apply deduces that it must apply at the
end of the file. ("context 2" is considered garbage and ignored.) The fix
is that removal lines must be counted towards the context of the old side.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Add a context menu item to invoke blame -C -C -C on a chunk
of the file. The results are used to update the 'original
location' column of the blame display.
The chunk is computed as the smallest line range that covers
both the 'last change' and 'original location' ranges of the
line that was clicked to open the menu.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Currently 'git-gui blame' does not kill its back-end
process, hoping that it will die anyway when the pipe
is closed. However, in some cases the process works
for a long time without producing any output. This
behavior results in a runaway CPU hog.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
On huge repositories, -C -C can be way too slow to be
unconditionally enabled, and it can also be useful to control
its precision.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Now that MERGE_RR was moved out of .git/rr-cache/, we have to delete
it somewhere else. Just in case somebody wants to use a newer git-gui
with an older Git, the file .git/rr-cache/MERGE_RR is removed, too (if
it exists).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
This adds a context menu entry below "Stage/Unstage Hunk" that stages or
unstages just the line under the mouse pointer.
This is by itself useful, for example, if there are unrelated changes in
the same hunk and the hunk cannot be split by reducing the context.
The feature can also be used to split a hunk by staging a number of
additions (or unstaging a number of removals) until there are enough
context lines that the hunk gets split.
The implementation reads the complete hunk that the line lives in, and
constructs a new hunk by picking existing context lines, removing unneeded
change lines and transforming other change lines to context lines. The
resulting hunk is fed through 'git apply' just like in the "Stage/Unstage
Hunk" case.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
In git-gui after clicking either on 'Create New Repository' or
'Open Existing Repository' the form elements aren't centered like
they are pretty much everywhere else in the app. At least when ran
on a mac, haven't checked on other platforms.
Using grid instead of pack seems to fix this.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we are deleting a local branch from refs/heads/ we need to
make sure any associated configuration stored in .git/config is
also removed (such as branch.$name.remote and branch.$name.merge).
The easiest way to do this is to use git-branch as that automatically
will look for and delete configuration keys as necessary.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
When creating new branches if branch.autosetupmerge is not set, or
is set to true or always and we have been given a remote tracking
branch as the starting point for a new branch we want to create the
necessary configuration options in .git/config for the new branch
so that a no argument git-pull on the command line pulls from the
remote repository's branch.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Apparently aspell 0.50 does not recognize "$$cr master" as a command,
but instead tries to offer suggestions for how to correctly spell
the word "cr". This is not quite what we are after when we want
the name of the current dictionary.
Instead of locking up git-gui waiting for a response that may never
come back from aspell we avoid sending this command if the binary
we have started claims to be before version 0.60.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
On startup, git-gui warns if there are many loose objects. It does so by
saying, e.g., that there are "approximately 768 loose objects". But isn't
"768" a very accurate number? Lets say "750", which (while still being a
very precise number) sounds much more like an estimation.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Peter Karlsson pointed out there is no value in translating the
string "Apple", as this is used as the dummy label for the Apple
menu on Mac OS X systems.
The Apple menu is actually not the menu with the Apple corporate
logo, but the menu next to it, which shows the name of the
application and is typically called the application menu. Most users
of git-gui see this menu titled as "Git Gui". The actual label of
this menu comes from our Info.plist file and cannot be specified
by any other means. Translating this string in the Tcl PO files
is not necessary.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The width of the commit message text area is currently hard-coded
to 75 characters. This value might be not optimal for some projects.
For instance users who would like to generate GNU-style ChangeLog
file from git commit message might prefer commit messages of width
no longer than 70 characters.
This patch adds a global and per repository option "Commit Message
Text Width", which could be used to change the width of the commit
message text area.
Signed-off-by: Adam Piątyszek <ediap@users.sourceforge.net>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
In several places, only the background colour is set to an explicit
value, sometimes even "white". This does not work well with dark
colour themes.
This patch tries to set the foreground colour to "black" in those
situations, where an explicit background colour is set without defining
any foreground colour.
Signed-off-by: Philipp A. Hartmann <ph@sorgh.de>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If the user tries to commit their changes without actually staging
anything we used to display an informational dialog suggesting they
first stage those changes, then retry the commit feature.
Unfortunately I broke this in aba15f7 ("Ensure error dialogs always
appear over all other windows") and failed to fix it in the paper
bag fix that came one day after it.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
When selecting a local working directory for a new repository or a
location to clone an existing repository into we now set the insert
point at the end of the selected path, allowing the user to type in
any additional parts of the path if they so desire.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If the main window is the only toplevel we have open then we
don't have a valid grab right now, so we need to assume the
best toplevel to use for the parent is ".".
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We really only support Aspell, so showing the compatibility line from
ispell is of little value to end users.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If the user has somehow managed to make us execute ispell instead
of aspell, even though our code is invoking aspell, and ispell is
not recognizing the aspell command line options we use to invoke
it then we don't want a giant usage message back from ispell.
Instead we show the ispell version number, letting the user know
we don't actually support that spell checker.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we feed a bad dictionary name to aspell on startup it may appear
to start (as Tcl found the executable in our $PATH) but it fails to
give us the version string. In such a case the close of the pipe
will report the exit status of the process (failure) and that is
an error in Tcl.
We now trap the subprocess failure and display the stderr message
from it, letting the user know why the failure is happening. We then
disable the spell checker, but keep our object instance so the user
can alter their preferred dictionary through the options dialog, and
possibly restart the spell checker.
I was also originally wrong to use "error" here for the display
of the problem to the user. I meant to use "error_popup", which
will open a message box and show the failure in a GUI context,
rather than killing git-gui and showing the message on the console.
Noticed by Ilari on #git.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we reconnect to the spellchecker there is no reason to resetup
the binding for button 3 on our text widget to show the suggestion
list (if available).
Plus, by moving it out of _connect and into init we can now break
out of _connect earlier if there is something wrong with the pipe,
for example if the dictionary we were asked to load is not valid.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Users may or may not be using aspell here. About the only thing
we are using that is aspell specific (and not supported by ispell
or an ispell variant) is some command line options when we start
up aspell, and a forced encoding of UTF-8. Both of these can be
corrected and/or cleaned up by users through an aspell wrapper
script, or through further improvements to git-gui. There is no
reason to require our translated strings to reference a specific
spell checker, especially if that spell checker implementation is
not very suitable for the language being translated.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we somehow managed to get our spellchecker instance created but
aspell wasn't startable we may not finish _connect and thus may
find one or more of our fields was not initialized in the instance.
If we have an instance but no version, there is no reason to show
a version to the user in our about dialog. We effectively have no
spellchecker available.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we are opening an error dialog we want it to appear above all of
the other windows, even those that we may have opened with a grab
to make the window modal. Failure to do so may allow an error
dialog to open up (and grab focus!) under an existing toplevel,
making the user think git-gui has frozen up and is unresponsive,
as they cannot get to the dialog.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
And markup one missing string for translation.
Signed-off-by: Christian Stimming <stimming@tuhh.de>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We don't need to fill this entire horizontal cavity, it looks really
bad on some platforms to stretch the widget out to fill the window.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We don't want the list length, we need the string length.
Found due to a bad " character discovered in the text and
Tcl throwing 'unmatched open quote in list'.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Many user friendly tools like word processors, email editors and web
browsers allow users to spell check the message they are writing
as they type it, making it easy to identify a common misspelling
of a word and correct it on the fly.
We now open a bi-directional pipe to Aspell and feed the message
text the user is editing off to the program about once every 300
milliseconds. This is frequent enough that the user sees the results
almost immediately, but is not so frequent as to cause significant
additional load on the system. If the user has modified the message
text during the last 300 milliseconds we delay until the next period,
ensuring that we avoid flooding the Aspell process with a lot of
text while the user is actively typing their message.
We wait to send the current message buffer to Aspell until the user
is at a word boundary, thus ensuring that we are not likely to ask
for misspelled word detection on a word that the user is actively
typing, as most words are misspelled when only partially typed,
even if the user has thus far typed it correctly.
Misspelled words are highlighted in red and are given an underline,
causing the word to stand out from the others in the buffer. This is
a very common user interface idiom for displaying misspelled words,
but differs from one platform to the next in slight variations.
For example the Mac OS X system prefers using a dashed red underline,
leaving the word in the original text color. Unfortunately the
control that Tk gives us over text display is not powerful enough
to handle such formatting so we have to work with the least common
denominator.
The top suggestions for a misspelling are saved in an array and
offered to the user when they right-click (or on the Mac ctrl-click)
a misspelled word. Selecting an entry from this menu will replace
the misspelling with the correction shown. Replacement is integrated
with the undo/redo stack so undoing a replacement will restore the
misspelled original text.
If Aspell could not be started during git-gui launch we silently eat
the error and run without spell checking support. This way users
who do not have Aspell in their $PATH can continue to use git-gui,
although they will not get the advanced spelling functionality.
If Aspell started successfully the version line and language are
shown in git-gui's about box, below the Tcl/Tk versions. This way
the user can verify the Aspell function has been activated.
If Aspell crashes while we are running we inform the user with an
error dialog and then disable Aspell entirely for the rest of this
git-gui session. This prevents us from fork-bombing the system
with Aspell instances that always crash when presented with the
current message text, should there be a bug in either Aspell or in
git-gui's output to it.
We escape all input lines with ^, as recommended by the Aspell manual
page, as this allows Aspell to properly ignore any input line that is
otherwise looking like a command (e.g. ! to enable terse output). By
using this escape however we need to correct all word offsets by -1 as
Aspell is apparently considering the ^ escape to be part of the line's
character count, but our Tk text widget obviously does not.
Available dictionaries are offered in the Options dialog, allowing
the user to select the language they want to spellcheck commit
messages with for the current repository, as well as the global
user setting that all repositories inherit.
Special thanks to Adam Flott for suggesting connecting git-gui
to Aspell for the purpose of spell checking the commit message,
and to Wincent Colaiuta for the idea to wait for a word boundary
before passing the message over for checking.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If the horizontal scrollbar isn't currently visible (because it has
not been needed) but we get an update to the scroll port we may find
the scrollbar window exists but the Tcl command doesn't. Apparently
it is possible for Tk to have partially destroyed the scrollbar by
removing the Tcl procedure name but still leaving the widget name in
the window registry.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The code we use to test if a hook is executable or not differs on
Cygwin from the normal POSIX case. Rather then repeating that for
all three hooks we call in our commit code path we can place the
common logic into a global procedure and invoke it when necessary.
This also lets us get rid of the ugly "|& cat" we were using before
as we can now rely on the Tcl 8.4 feature of "2>@1" or fallback to
the "|& cat" when necessary.
The post-commit hook is now run through the same API, but its outcome
does not influence the commit status. As a result we now show any of
the errors from the post-commit hook in a dialog window, instead of on
the user's tty that was used to launch git-gui. This resolves a long
standing bug related to not getting errors out of the post-commit hook
when launched under git-gui.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
During i18n translation work this message was partially broken
by using "append" instead of "strcat" to join the two different
parts of the message together.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Under core Git the git-commit tool will invoke the commit-msg hook
if it exists and is executable to the user running git-commit. As
a hook it has some limited value as it cannot alter the commit, but
it can modify the message the user is attempting to commit. It is
also able to examine the message to ensure it conforms to some local
standards/conventions.
Since the hook takes the name of a temporary file holding the message
as its only parameter we need to move the code that creates the temp
file up earlier in our commit code path, and then pass through that
file name to the latest stage (where we call git-commit-tree). We let
the hook alter the file as it sees fit and we don't bother to look at
its content again until the commit succeeded and we need the subject
for the reflog update.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we apply a hunk in either direction this may change the file's
status. For example if a file is completely unstaged, and has at
least two hunks in it and the user stages one hunk the file will
change from "Modified, not staged" to "Portions staged for commit".
Resetting the file path causes our trace on this variable to fire;
that trace is used to update the file header in the diff viewer to
the file's current status.
Noticed by Johannes Sixt.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Often users setup a few source files and get a project rolling
before they create a Git repository for it. In such cases the
core Git tools allow users to initialize a new repository by
simply running `git init` at the desired root level directory.
We need to allow the same situation in git-gui; if the user is
trying to make a new repository we should let them do that to any
location they chose. If the directory already exists and already
has files contained within it we still should allow the user to
create a repository there. However we still need to disallow
creating a repository on top of an existing repository.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
I broke this extremely cool feature in 1ac17950,
but it is rather easy to fix this. Sorry for that.
Signed-off-by: Christian Stimming <stimming@tuhh.de>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Concatenating commit log messages from multiple commits works better
when all of the commits end with a clean line break.
Its good to be strict in what you create, and lenient in what you
accept, and since we're creating here, we should always try to
Do The Right Thing(tm).
Signed-off-by: Bernt Hansen <bernt@alumni.uwaterloo.ca>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Johannes Sixt pointed out the diff headers "old mode ..." and
"new mode ..." were not being parsed properly by git-gui. We
now include them in the diff viewer for a file.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Previusly, there was only a 16x16 image, which looked very distorted.
Here we add a 32x32 version, and also make the image sharper.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
I'm seeing a lot of silent failures from git-update-index on
Windows and this is leaving the index.lock file intact, which
means users are later unable to perform additional operations.
When the index is locked behind our back and we are unable to
use it we may need to allow the user to delete the index lock
and try again. However our UI state is probably not currect
as we have assumed that some changes were applied but none of
them actually did. A rescan is the easiest (in code anyway)
solution to correct our UI to show what the index really has
(or doesn't have).
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Git progress bars from tools like git-push and git-fetch use CR
to skip back to the start of the current line and redraw it with
an updated progress. We were doing this in our Tk widget but had
failed to skip the CR, which Tk doesn't draw well.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Post Git 1.5.3 a new style progress bar has been introduced that
uses only one line rather than two. The formatting of the completed
and total section is also slightly different so we must adjust our
regexp to match. Unfortunately both styles are in active use by
different versions of Git so we need to look for both.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If git-write-tree fails (such as if the index file is currently
locked and it wants to write to it) we were not getting the error
message as $tree_id was always the empty string so we shortcut
through the catch and never got the output from stderr.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Apparently native Tcl/Tk on Windows is using \ as the return value
from [file separator] but [file normalize] on that same system is
using / rather than \ to represent a directory separator. I really
think that is nuts, but its what is happening.
So we can actually just hardcode our separator to / as all systems
we support (Windows, Mac OS X, UNIX) use /.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If the user tries to clone a Git repository that is actually a
workdir of another repository (by way of contrib git-new-workdir)
then the contents of .git is a series of Windows .lnk files which
Tcl can't read if this is a native Tcl process. To read the real
objects directory we need to resolve the link to that location.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
On Windows its better to use a shortcut (.lnk file) over a batch
script (.bat) as we can specify the icon file for the .lnk and
thus have these git specific objects appear on the desktop with
that git specific icon file.
Unfortunately the authors of Tcl did not bless us with the APIs
needed to create shortcuts from within Tcl. But Microsoft did
give us Windows Scripting Host which allows us to execute some
JavaScript that calls some sort of COM object that can operate
on a .lnk file.
We now build both Cygwin and non-Cygwin "desktop icons" as proper
Windows .lnk files, using the "Start in" property of these files
to indicate the working directory of the repository the user wants
to launch.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
On Windows we need to actually setup binds for the accelerator
keys, otherwise the OS doesn't respond to them when the user
presses the key combinations. Apparently we automatically get
these on Mac OS X when we configure the menu commands, but not
on Windows.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Johannes Sixt