[RFC] Off-by-one in linkmap cairo drawing ?
|
View:
New views
6 Messages
—
Rating Filter:
Alert me
|
 |
[RFC] Off-by-one in linkmap cairo drawing ?
by Vincent Legoll
::
Rate this Message:
Hello,
could someone please comment on the validity of: Attachment #133349: maybe this one is more complete for bug #580311 The screenshots were taken with the first (obsoleted) patch What are all those magical 0.5 values for anyways ? -- Vincent Legoll [fix-cairo-linkmap-drawing2.patch] diff --git a/filediff.py b/filediff.py index ac5063c..cf2979f 100644 --- a/filediff.py +++ b/filediff.py @@ -1116,8 +1116,10 @@ class FileDiff(melddoc.MeldDoc, gnomeglade.Component): return [self._pixel_to_line(idx, pix_start[idx]), self._pixel_to_line(idx, pix_start[idx]+htotal)] visible = [None] + bounds(which) + bounds(which+1) + adjust = 0 + # For bezier control points - x_steps = [-0.5, (1. / 3) * wtotal, (2. / 3) * wtotal, wtotal + 0.5] + x_steps = [-adjust, (1. / 3) * wtotal, (2. / 3) * wtotal, wtotal + adjust] def paint_pixbuf_at(pixbuf, x, y): context.translate(x, y) @@ -1148,14 +1150,14 @@ class FileDiff(melddoc.MeldDoc, gnomeglade.Component): if t0 == t1: t0 -= 1 - context.move_to(x_steps[0], f0 - 0.5) - context.curve_to(x_steps[1], f0 - 0.5, - x_steps[2], t0 - 0.5, - x_steps[3], t0 - 0.5) - context.line_to(x_steps[3], t1 + 0.5) - context.curve_to(x_steps[2], t1 + 0.5, - x_steps[1], f1 + 0.5, - x_steps[0], f1 + 0.5) + context.move_to(x_steps[0], f0 - adjust) + context.curve_to(x_steps[1], f0 - adjust, + x_steps[2], t0 - adjust, + x_steps[3], t0 - adjust) + context.line_to(x_steps[3], t1 + adjust) + context.curve_to(x_steps[2], t1 + adjust, + x_steps[1], f1 + adjust, + x_steps[0], f1 + adjust) context.close_path() context.set_source_rgb(*self.fill_colors[c[0]]) @@ -1171,8 +1173,8 @@ class FileDiff(melddoc.MeldDoc, gnomeglade.Component): paint_pixbuf_at(pix0, 0, f0) # allow for scrollbar at end of textview - mid = int(0.5 * self.textview[0].allocation.height) + 0.5 - context.set_source_rgba(0., 0., 0., 0.5) + mid = int(adjust * self.textview[0].allocation.height) + adjust + context.set_source_rgba(0., 0., 0., adjust) context.move_to(.35 * wtotal, mid) context.line_to(.65 * wtotal, mid) context.stroke() _______________________________________________ meld-list mailing list meld-list@... http://mail.gnome.org/mailman/listinfo/meld-list |
|
|
Re: [RFC] Off-by-one in linkmap cairo drawing ?
by Stephen Kennedy
::
Rate this Message:
I think they may have been due to some strangeness in the way
cairo did antialiasing. See the original commit http://git.gnome.org/cgit/meld/commit/?id=ca289a59a7e3a4dc690ecf7aa7606700dc246835 I didn't track down where 0.2 became 0.5 That was quite tedious to find using cgit and gitk. Does anybody know is it not possible with git to jump directly to the diff which corresponds to a given line of a given revision? e.g. like: http://svn.gnome.org/viewvc/meld/trunk/filediff.py?annotate=1331&pathrev=1333 Stephen. On Sun, Apr 26, 2009 at 6:22 PM, Vincent Legoll <vincent.legoll@...> wrote: > Hello, > > could someone please comment on the validity of: > > Attachment #133349: maybe this one is more complete > for bug #580311 > > The screenshots were taken with the first (obsoleted) patch > > What are all those magical 0.5 values for anyways ? > > -- > Vincent Legoll > > _______________________________________________ > meld-list mailing list > meld-list@... > http://mail.gnome.org/mailman/listinfo/meld-list > meld-list mailing list meld-list@... http://mail.gnome.org/mailman/listinfo/meld-list |
|
|
|
|
|
Re: [RFC] Off-by-one in linkmap cairo drawing ?
by Kai Willadsen-2
::
Rate this Message:
2009/4/27 Vincent Legoll <vincent.legoll@...>:
> Hello, > > could someone please comment on the validity of: > > Attachment #133349: maybe this one is more complete > for bug #580311 > > The screenshots were taken with the first (obsoleted) patch The horizontal line in your after-screenshots is blurred, which is what the 0.5 adjustment fixes. > What are all those magical 0.5 values for anyways ? Rough answer: integer coordinates in Cairo fall between pixels, not on pixels Longer answer: http://cairographics.org/FAQ/#sharp_lines I changed the 0.2 to 0.5 a while back since the 0.2 was only there to fix blurry rendering of straight lines, and for that purpose 0.5 is what should actually be used. As for the bug, the 'missing pixel' was the intended result at the time. The idea is to indicate the insertion point of the diff, while maintaining nice sharp straight lines where appropriate. I agree that our rendering of these lines could be nicer... but if you play around with different cases it quickly becomes non-trivial to get it right. We should really decide how the lines *should* be placed, and then figure out the rendering. Kai _______________________________________________ meld-list mailing list meld-list@... http://mail.gnome.org/mailman/listinfo/meld-list |
|
|
history digging with git (was: [RFC] Off-by-one in linkmap cairo drawing ?)
by Martin Renold
::
Rate this Message:
hi Stephen,
On Tue, Apr 28, 2009 at 09:15:32AM +0100, Stephen Kennedy wrote: > That was quite tedious to find using cgit and gitk. Does anybody > know is it not possible with git to jump directly to the diff which > corresponds to a given line of a given revision? e.g. like: > http://svn.gnome.org/viewvc/meld/trunk/filediff.py?annotate=1331&pathrev=1333 I was looking for the same thing for some time now, and I have not found a single program that does it. But I discovered something even better! Git allows you to grep through all diffs. In your example, I saw the comment "bezier" above the code of interest, and, using this as an "anchor keyword", git log -p -Sbezier finds just the change you were looking for. It shows both commit message and patch, and is quite fast. This command opens a tiny new paradise for history digging. I'm finally starting to enjoy git... bye, Martin _______________________________________________ meld-list mailing list meld-list@... http://mail.gnome.org/mailman/listinfo/meld-list |
|
|
Re: history digging with git (was: [RFC] Off-by-one in linkmap cairo drawing ?)
by Stephen Kennedy
::
Rate this Message:
On Wed, May 13, 2009 at 9:49 PM, Martin Renold <martinxyz@...> wrote:
> On Tue, Apr 28, 2009 at 09:15:32AM +0100, Stephen Kennedy wrote: >> That was quite tedious to find using cgit and gitk. Does anybody >> know is it not possible with git to jump directly to the diff which >> corresponds to a given line of a given revision? e.g. like: >> http://svn.gnome.org/viewvc/meld/trunk/filediff.py?annotate=1331&pathrev=1333 > > I was looking for the same thing for some time now, and I have not found a > single program that does it. But I discovered something even better! > > Git allows you to grep through all diffs. In your example, I saw the comment > "bezier" above the code of interest, and, using this as an "anchor keyword", > > git log -p -Sbezier Thanks Martin, that's a good trick to know. I recently found that qgit can do a reasonable job. Turn on the tree view (T), then you get an annotated file history by double clicking a file. Double clicking a line goes to the commit which introduced that line and you can view it with (Ctrl+P). Stephen. _______________________________________________ meld-list mailing list meld-list@... http://mail.gnome.org/mailman/listinfo/meld-list |
| Free embeddable forum powered by Nabble | Forum Help |
