Now that we have a general-purpose way of taking some action when a
commit ID of interest is encountered, use that for triggering the
git diff-index process when we find the currently checked-out head,
rather than the special-purpose lookingforhead variable.

Also do the commitinterest processing in getcommitlines rather than
in showstuff.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

We weren't updating the rowfinal list in insertrow and removerow, so
it was getting out of sync with rowidlist, which resulted in Tcl errors.
This also optimizes the setting of rowfinal in layoutrows a bit.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This arranges things so that we can do the layout all the way up to
the last commit that we have received from git log.  If we get more
commits we re-lay and redisplay (if necessary) the visible rows.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This adds code to write out the topology information used to determine
precedes/follows and branch information into a cache file (~3.5MB for
the kernel tree).  At startup we read the cache file and then do a
git rev-list to update it, which is fast because we exclude all commits
in the cache that have no children and commits reachable from them
(which amounts to everything in the cache).  If one of those commits
without children no longer exists, then git rev-list will give an error,
whereupon we throw away the cache and read in the whole tree again.

This gives a significant speedup in the startup time for gitk.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

When "Show nearby tags" is turned off and the user did a cherry-pick,
we were trying to access variables relating to the descendent/ancestor
tag & head computations in addnewchild though they hadn't been set.
This makes sure we don't do that.  Reported by Johannes Sixt.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

If the view we're switching to hadn't been read in, we hit an early
return in showview which meant we didn't update the ref list window.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This changes layoutrows and optimize_rows to make it possible to lay
out only a little bit more of the graph than is visible, rather than
having to lay out the whole graph from top to bottom.  To lay out
some of the graph without starting at the top, we use the new make_idlist
procedure for the first row, then lay it out proceeding downwards
as before.  Empty list elements in rowidlist are used to denote rows
that haven't been laid out yet.

Optimizing happens much as before except that we don't try to optimize
unless we have three consecutive rows laid out (or the top 2 rows).
We have a new list, rowisopt, to record which rows have been optimized.

If we change a row that has already been drawn, we set a flag which
causes drawcommits to throw away everything drawn on the canvas and redraw
the visible rows.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Instead, when looking for lines that should be terminated with a down
arrow, we look at the parents of the commit $downarrowlen + 1 rows
before.  This gets rid of one more place where we are assuming that
all the rows are laid out in order from top to bottom.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

First, this fixes the problem where a SHA1 id wouldn't be displayed
as a link if it wasn't in the part of the graph that had been laid
out at the time the details pane was filled in, even if that commit
later became part of the graph.  This arranges for us to turn the
SHA1 id into a link when we get to that id in laying out the graph.

Secondly, there was a problem where the cursor wouldn't always turn
to a hand when over a link, because the areas for two links could
overlap slightly.  This fixes that by using a counter rather than
always reverting to a counter when we leave the region of a link
(which can happen just after we've entered a different link).
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This changes layoutrows to use information from rowidlist and children
to work out which parent ids are appearing for the first time or need
an up arrow, instead of using idinlist.  To detect the situation where
git log doesn't give us all the commits it references, this adds an
idpending array that is updated and used by getcommitlines.

This also fixes a bug where we weren't resetting the ordertok array when
updating the list of commits; this fixes that too, and a bug where we
could try to access an undefined element of commitrow if the user did
an update before gitk had finished reading in the graph.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Instead make the rowranges procedure compute its result by looking
in the rowidlist entries for the rows around the children of the id
and the id itself.  This turns out not to take too long, and not having
to maintain idrowranges and rowrangelist speeds up the layout.

This also makes optimize_rows not use rowranges, since all it really
needed was a way to work out if one id is the first child of another,
so it can just look at the children list.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This adds an entry to the File menu labelled "List references" which
pops up a window showing a sorted list of branches, tags, and other
references, with a little icon beside each to indicate what sort it
is.  The list only shows refs that point to a commit that is included
in the graph, and if you click on a ref, the corresponding commit
is selected in the main window.  The list of refs gets updated
dynamically.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

If a commit contained a copy operation, the file name was not correctly
determined, and the corresponding part of the patch could not be
navigated to from the list of files.
Signed-off-by: NJohannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

My fix in commit b1054ac9 was only
half-right, since it ignored the case where the descendent heads of
the head being removed correspond to two or more different commits.
This fixes it.  Reported by Mark Levedahl.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The new 'datetimeformat' configuration variable in ~/.gitk can be set
to a Tcl 'clock format' format string to modify the display of dates
and times.

http://www.tcl.tk/man/tcl8.4/TclCmd/clock.htm has a list of allowed
fields.
Signed-off-by: NArjen Laarhoven <arjen@yaph.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

More lines of context sometimes help to better understand a diff.
This patch introduces a text field above the box displaying the
blobdiffs. You can type in the number of lines of context that
you wish to view. The number of lines of context is saved to
~/.gitk.
Signed-off-by: NSteffen Prohaska <prohaska@zib.de>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

When we had two heads on the same commit, and the user tried to remove
one of them, gitk was sometimes incorrectly saying that the commits
on that branch weren't on any other branch.  This fixes it.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

If "Show nearby tags" is turned off, selecting "Update" from the File
menu will cause a Tcl error.  This fixes it.  The problem was that
we were calling regetallcommits unconditionally, but it assumed that
getallcommits had been called previously.  This also restructures
{re,}getallcommits to be a bit simpler.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

In fixing the "can't unset idinlist" error, I moved the setting of
idinlist into the loop that splits the parents into "new" parents
(i.e. those of which this is the first child) and "old" parents.
Unfortunately this is incorrect in the case where we hit the break
statement a few lines further down, since when we come back in,
we'll see idinlist($p) set for some parents that aren't in the list.

This fixes it by moving the loop that sets up newolds and oldolds
further down.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This is to help people starting gitk from graphical file managers where
the stderr output is hidden.
Signed-off-by: NAlex Riesen <raa.lkml@gmail.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

If there is no commit made yet, gitk just dumps a Tcl error on stderr,
which sometimes is hard to see.  Noticed when gitk was run from Xfce
file manager (thunar's custom action).
Signed-off-by: NAlex Riesen <raa.lkml@gmail.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Windows, unlike X-Windows, sends mousewheel events by default to the
window that has keyboard focus and uses the MouseWheel event to do so.
The window to be scrolled must be able to take focus, but gitk's panels
are disabled so cannot take focus.  For all these reasons, a different
design is needed to use the mousewheel on Windows.  The approach here is
to bind the mousewheel events to the top level window and redirect them
based upon the current mouse position.
Signed-off-by: NMark Levedahl <mdl123@verizon.net>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

On windows, mouse input follows the keyboard focus, so to allow selecting
text from the patch canvas we must not shift focus back to the top level.
This change has no negative impact on X, so we don't explicitly test
for Win32 on this change. This provides similar selection capability
as already available using X-Windows.
Signed-off-by: NMark Levedahl <mdl123@verizon.net>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Under some circumstances, having duplicate parents in a commit could
trigger a "can't unset idinlist" Tcl error.  This fixes the cause
(the logic in layoutrows could end up putting the same commit into
rowidlist twice) and also puts a catch around the unset to ignore
the error.

Thanks to Jeff King for coming up with a test script to generate a
repo that shows the problem.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This changes the optimizer to insert pads to straighten downward
pointing arrows so they point straight down.  When drawing the parent
link to the first child in drawlineseg, this draws it with 3 segments
like other parent links if it is only one row high with an arrow.
These two things mean we can dispense with the workarounds for arrows
on diagonal segments.  This also fixes a couple of other minor bugs.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The way gitk used to draw the lines joining a commit to the lines
representing its parents was sometimes visually ambiguous, especially
when the line to the parent had a corner that coincided with a corner
on another line.

This improves things by using a smaller slanting section on the line
joining a commit to a parent line if the parent line is vertical where
it joins on.  It also optimizes the drawing a little in the case where
the parent line slants towards this commit already.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This creates an "ordering token" for each commit which establishes
a total ordering for commits and is used to order the commits from
left to right on a row.  The ordering token is assigned when a commit
is first encountered or when it is first listed as a parent of some
other commit, whichever comes first.  The ordering token is a string
of variable length.  Parents that don't already have an ordering
token are assigned one by appending to the child's token; the first
parent gets a "0" on the end, the second "1" and so on.  As an
optimization, the "0" isn't appended if the child only has one parent.

When inserting a new commit into an element of rowidlist, it is
inserted in the position which makes the ordering tokens increase
from left to right.

This also simplifies the layout code by getting rid of the rowoffsets
variable, and terminates lines with an arrow after 5 rows if the line
would be longer than about 110 rows (rather than letting them go on
and terminating them later with an arrow if the graph gets too wide).

The effect of having the total ordering, and terminating the lines
early, is that it will be possible to lay out only a part of the graph
rather than having to do the whole thing top to bottom.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

At the moment this just has two entries, which allow you to add the file
that you clicked on to the list of filenames to highlight, or replace
the list with the file.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This fixes the problem reported by Brian Downing where searching for
a string that doesn't exist would give a Tcl error.  The basic problem
was that we weren't reading the data for the last commit since it
wasn't terminated with a null.  This effectively adds a null on the end
(if there isn't one already) to make sure we process the last commit.

This also makes the yellow background behind instances of the search
string appear more consistently, and fixes a bug where the "/" key
would just find the same commit again and again instead of advancing.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

When the git log process returned an error immediately, we were
sometimes getting no main window and no error window displayed,
with the gitk process just hanging waiting for something.  It appears
that the tkwait in show_error, which waits for the error window to
be destroyed, wasn't sufficient to allow the main window or the error
window to be mapped.

This adds a wait in the main startup code after the main window
has been created to wait until it is visible.  This seems to fix the
problem.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

git-gui already uses the command key for accelerators, but gitk has
never done so.  I'm actually finding it very hard to move back and
forth between the two applications as git-gui is following the Mac
OS X conventions and gitk is not.

This trick is the same one that git-gui uses to determine which
key to bind actions to.
Signed-off-by: NShawn O. Pearce <spearce@spearce.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Cygwin's Tcl is configured to honor any occurence of ctrl-z as an
end-of-file marker, while some commits in the git repository and possibly
elsewhere include that character in the commit comment. This causes gitk
ignore commit history following such a comment and incorrect graphs. This
change affects only Windows as Tcl on other platforms already has
eofchar == {}. This fixes problems noted by me and by Ray Lehtiniemi, and
the fix was suggested by Shawn Pierce.
Signed-off-by: NMark Levedahl <mdl123@verizon.net>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The magenta was a bit close in color to the normal blue commits.  This
makes them green instead as suggested by Linus.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This makes gitk show up to two fake commits when there are local changes
in the repository; one to represent the state of the index and one to
represent the state of the working directory.  The commit representing
the working directory is colored red as before; the commit representing
the index state is colored magenta (as being between red and blue in
some sense).
Signed-off-by: NPaul Mackerras <paulus@samba.org>

When I added the "--" case to the code scanning the arguments, I missed
the fact that since the switch statement uses -regexp, the "--" case
will match any argument containing "--", e.g. "--all".  This fixes it
by taking out the -regexp (since we don't actually need regular
expression matching) and adjusting the match strings.

A side effect of this is that previously any argument starting with
"-d" would be taken to indicate date mode; now the argument has to be
exactly "-d" if you want date mode.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This makes gitk more consistent with git rev-list and git log in its
handling of arguments that could be either a revision or a filename;
now gitk displays an error message and quits, rather than treating it
as a revision and getting an error in the underlying git log.  Now
gitk always passes "--" to git log even if no filenames are being
specified.

It also makes gitk display errors in invoking git log in a window
rather than on stderr, and makes gitk stop looking for a -d flag
when it sees a "--" argument.
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This is based on patches from Linus Torvalds and Junio Hamano, so the
ideas here are theirs.

This makes gitk use "git log -z --pretty=raw" instead of "git rev-list"
to generate the list of commits, and also makes it grok the "<" and ">"
markers that git log (and git rev-list) output with the --left-right
flag to indicate which side of a symmetric diff a commit is reachable
from.  Left-side commits are drawn with a triangle pointing leftwards
instead of a circle, and right-side commits are drawn with a triangle
pointing rightwards.  The commitlisted list is used to store the
left/right information as well as the information about whether each
commit is on the boundary.
Signed-off-by: NPaul Mackerras <paulus@samba.org>