Re: [Quilt-dev] [PATCH] Enhanced decoration for "series -v" command

[Peter Williams]

Peter Williams wrote:
> Andreas Gruenbacher wrote:
>
> > On Friday 10 June 2005 08:21, Peter Williams wrote:
> >
> > > I am the author of a PyGTK GUI wrapper ( see
> > > <http://freshmeat.net/projects/gquilt/>) for quilt and it has been
> > > suggested to me that the panel containing the display of the patch
> > > series could be enhanced by having it display (for applied patches)
> > > whether each patch needs refreshing or not.
> >
> >
> >
> > Looks nice. It seems the Push and Pop icons point in the wrong 
> > direction wrt. the series file (which has its top at the bottom)?
>
>
> You're the second person to say that and I don't understand as the push 
> arrow pushes up onto the bottom of the series and the pop arrow pops 
> down off the bottom of the series (at least in the latest versions). 
> Older versions used stock icons with unpredictable results depending on 
> how they were represented on each system so I've gone to gquilt specific 
> icons for later versions.  Can you confirm that you've got version 0.4?
>
> > > The attached patch modifies quilt's "series -v" command to add this
> > > information to the output.
> >
> >
> >
> > files_may_have_changed may return false positives, so the resulting 
> > status information isn't worth much. Also it doesn't take care of 
> > shadowed files, so modifying a file in a patch further down in the 
> > series file will mark all previous patches that include that file as 
> > modified.
>
>
> Yes, I've discovered that :-( and will try to come up with a fix in a 
> day or two.  Basically, when a file is newer than the patch's timestamp 
> I'll look at applied patches above the patch until I find one that 
> shares the same file and make the comparison with the "cached" file for 
> that patch.
>
> As for other false positives I'm not sure that they're worth hunting 
> down as they should only be caused (I think) by rare occurrences such as 
> a change being made to a file and then removed without a refresh being 
> done.  I'm reluctant to go to all the trouble of doing diffs to confirm 
> that there is indeed a difference in the file (due to the overhead) but 
> will do that if necessary.

Attached is a patch (on top of my previous patch) that addresses the 
false positives issue.  It seems to work well with one of my playgrounds 
that has quite extensive overlapping of patches.

The reason that I've provided this as a patch on top of my previous 
patch is to make it easier to see the changes involved.

I'm fairly sure that this change won't have any adverse effects for the 
use of files_may_have_changed() within the "pop" command but would value 
the opinion of others.

Peter
--
Peter Williams                                   address@hidden

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce

Index: quilt/scripts/patchfns.in
===================================================================
--- quilt.orig/scripts/patchfns.in
+++ quilt/scripts/patchfns.in
@@ -340,7 +340,15 @@ files_may_have_changed()
        local apply_ts=$(date -r "$QUILT_PC/$patch/.timestamp" '+%s') ts
        for file in $(files_in_patch $patch)
        do
-               ts=$(date -r $file '+%s' 2> /dev/null)
+               local next_patch=$(next_patch_for_file $patch $file) test_file
+
+               if [ -z "$next_patch" ]
+               then
+                       test_file=$file
+               else
+                       test_file=$(backup_file_name $next_patch $file)
+               fi
+               ts=$(date -r $test_file '+%s' 2> /dev/null)
                [ -z "$ts" ] && return 0
                [ "$ts" -gt $apply_ts ] && return 0
        done
Index: quilt/quilt/refresh.in
===================================================================
--- quilt.orig/quilt/refresh.in
+++ quilt/quilt/refresh.in
@@ -281,6 +281,7 @@ if [ -e $patch_file ] && \
     @DIFF@ -q $patch_file $tmp_result > /dev/null
 then
        printf $"Patch %s is unchanged\n" "$(print_patch $patch)"
+       touch $QUILT_PC/$patch/.timestamp
        exit 0
 fi