Re: [RFC/PATCHv2 1/5] worktree: provide better prefix to go back to original cwd

[Nguyen Thai Ngoc Duy <pclouds@gmail.com>]

On Sat, Oct 16, 2010 at 07:48:02PM -0700, Chris Packham wrote:
> On 16/10/10 11:42, Jonathan Nieder wrote:
> > Hi,
> > 
> > Chris Packham wrote:
> > 
> >> From: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> >>
> >> When both GIT_DIR and GIT_WORK_TREE are set, if cwd is outside worktree,
> >> prefix (the one passed to every builtin commands) will be set to NULL,
> >> which means "user stays at worktree topdir".
> >>
> >> As a consequence, command line arguments are supposed to be relative
> >> to worktree topdir, not current working directory. Not very intuitive.
> > 
> > Thanks.  More detailed history for this patch:
> > 
> >  - v0: http://thread.gmane.org/gmane.comp.version-control.git/157599/focus=157601
> >  - v1: http://thread.gmane.org/gmane.comp.version-control.git/158287
> >  - v2: http://thread.gmane.org/gmane.comp.version-control.git/158369
> > 
> 
> I think I must have missed v2. I was playing around with my gmail
> filters around that time so I could have missed them. Actually now I've
> found the message it's missing the last 'm' in gmail.com. I'll grab the
> latest patch and give it a test when I get a chance.

I missed the last "m" in your email address. That's why v2 never reached you.
I thought I sent you an email but probably forgot it.

Anyway v2 does not work if worktree and cwd are on different Windows drives.
This on top should fix it:

---8<---
diff --git a/setup.c b/setup.c
index 2389a9e..35d2691 100644
--- a/setup.c
+++ b/setup.c
@@ -371,12 +371,8 @@ static const char *setup_prefix(const char *cwd)
 	}
 	/* get /foo/, not /foo/baa if /foo/baa1 and /foo/baa2 are given */
 	else if (worktree[len] && cwd[len]) {
-		while (len && !is_dir_sep(worktree[len]))
-			len--;
-		len++;
-
 		/* Worktree and cwd are on different drives? */
-		if (len == 3 && has_dos_drive_prefix(cwd)) {
+		if (!len && has_dos_drive_prefix(cwd)) {
 			if (startup_info) {
 				/* make_path_to_path will add the trailing slash */
 				startup_info->cwd_to_worktree = make_path_to_path(NULL, worktree);
@@ -384,6 +380,10 @@ static const char *setup_prefix(const char *cwd)
 			}
 			return NULL;
 		}
+
+		while (len && !is_dir_sep(worktree[len]))
+			len--;
+		len++;
 	}
 	else {
 		if (worktree[len]) {
---8<---

> 
> > Any thoughts about the previous questions?
> > 
> 
> I haven't caught up on the newest thread so no great revelations. Except
> that for the grep submodules use-case we can assume that the worktree
> will be a subdirectory of the cwd. I don't think we want to limit
> ourselves to that one use-case.

While at it, have you thought of support --recursive and
--full-tree [1]? There are issues with --full-tree and prefixes [2],
which is why it is dropped but I think it's a good idea.

--full-tree disregards where you stand and greps in whole repo. In a
repo with submodules, that would mean grep the supermodule and all
submodules regardless where you stand, even if you stand in a
submodule.

[1] http://mid.gmane.org/7vk4xggv27.fsf@alter.siamese.dyndns.org
[2] http://mid.gmane.org/7vskaqptvj.fsf@alter.siamese.dyndns.org
-- 
Duy