[PATCH] subrepo: avoids empty commit when .hgsubstate is dirty (issue2403)

[Erik Zielke]

fre, 17 12 2010 kl. 03:38 +0100, skrev Mads Kiilerich: 
> Erik Zielke wrote, On 11/29/2010 09:38 AM:
> > # HG changeset patch
> > # User Erik Zielke<ez at aragost.com>
> > # Date 1291019843 -3600
> > # Node ID b1bcf90cb59e3c92708acd5d894ed65d8a3cd30b
> > # Parent  da69a1597285fe25eeedab1a45869487773b715b
> > subrepo: avoids empty commit when .hgsubstate is dirty (issue2403)
> 
> Combining an (otherwise) ignored dirty .hgsubstate still gives some 
> strange results:
> 
> diff --git a/tests/test-subrepo-empty-commit.t 
> b/tests/test-subrepo-empty-commit.t
> --- a/tests/test-subrepo-empty-commit.t
> +++ b/tests/test-subrepo-empty-commit.t
> @@ -28,6 +28,34 @@
>     nothing changed
>     [1]
> 
> +but this looks suspicious:
> +
> +  $ hg id sub
> +  5bbc614a5b06 tip
> +  $ echo c1b > f1
> +  $ hg commit -m1
> +  committing subrepository sub
> +  $ hg log -p -r tip
> +  changeset:   1:c377c139cf3c
> +  tag:         tip
> +  user:        test
> +  date:        Thu Jan 01 00:00:00 1970 +0000
> +  summary:     1
> +
> +  diff -r 853ea21970bb -r c377c139cf3c .hgsubstate
> +  --- a/.hgsubstate    Thu Jan 01 00:00:00 1970 +0000
> +  +++ b/.hgsubstate    Thu Jan 01 00:00:00 1970 +0000
> +  @@ -1,1 +1,1 @@
> +  -5bbc614a5b06ad7f3bf7c2463d74b005324f34c1 sub
> +  +0000000000000000000000000000000000000000 sub
> +  diff -r 853ea21970bb -r c377c139cf3c f1
> +  --- a/f1    Thu Jan 01 00:00:00 1970 +0000
> +  +++ b/f1    Thu Jan 01 00:00:00 1970 +0000
> +  @@ -1,1 +1,1 @@
> +  -c1
> +  +c1b
> +
> +
>   an okay update of .hgsubstate
>     $ cd sub
>     $ echo c3 > f2
> @@ -49,7 +77,7 @@
>     committing subrepository sub
>     $ echo '0000000000000000000000000000000000000000 sub' > .hgsubstate
>     $ hg diff --nodates
> -  diff -r 41e1dee3d5d9 .hgsubstate
> +  diff -r b52e776d6fc4 .hgsubstate
>     --- a/.hgsubstate
>     +++ b/.hgsubstate
>     @@ -1,1 +1,1 @@
> 
> 
> Is that correct and intended?
> 

No, its not, should I just resend the patch with fixes or is the
procedure different when a patch is queued?

> /Mads