Thread (5 messages) flat view 5 messages, 4 authors, 2016-06-15

Re: [PATCH] '%S' option for pretty printing to support --source

From: Deskin Miller <hidden>
Date: 2016-06-15 22:46:20

On Thu, Mar 5, 2009 at 04:17, Jeff King [off-list ref] wrote:
On Thu, Mar 05, 2009 at 09:18:28AM +0200, Petri Hodju wrote:
quoted
+static void format_source(struct strbuf *sb, const struct commit *commit)
+{
+    if (commit->util)
+     strbuf_addstr(sb, (char *) commit->util);
+}
+
Hmm. This is the second patch in the last few weeks to use commit->util
to carry information for --pretty=format: (I am cc'ing Deskin Miller,
who wrote the first).
Thanks Jeff.  Fortunately I managed to catch this one anyway.

Petri, the patch series from me which Jeff is referring to is viewable at

http://thread.gmane.org/gmane.comp.version-control.git/111524/

for reference.

I am in the middle of a move and ought to be packing right now, so
needless to say my git budget at the moment is pretty much nil, and
will be so for at least another week I'd guess.  This is to say, I've
not done any additional work in light of Jeff's or Dscho's comments on
my series, though I intend to once I'm relocated.
They cannot both work, obviously. So we need to do one of:

 - refactor the information out of commit->util to somewhere else

 - allow multiple commit->util users somehow (which I think is a
   potential performance problem -- the simplistic design is meant to
   avoid things like allocation overhead)
I'm inclined to do as Dscho suggests here: glancing at the current
struct decoration usage briefly I think my reflog printing could work
that way with no problem.  However, this would largely ignore your
other comments about prettifying the pretty-printing code.  If a new
series using struct decoration isn't useful, let me know, otherwise
I'll plan on doing this once I have a chance.
 - gracefully block concurrent use of conflicting features
I agree that any blocking should be graceful, but ultimately I find
the idea of disallowing features because they happen to use the same
underlying implementation distasteful.  With a little work we should
be able to allow both with no problem.

Deskin Miller
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help