Re: [PATCH] maint: fix alphabetical order in .gitignore

[Bernhard Voelker]

On 01/11/2013 01:26 PM, Pádraig Brady wrote:
> One issue is if there were existing duplicate entries
> in .gitignore, then the counts would be inaccurate.
> That might be guarded with something like:
> 
>    duplicate_entries=$(sort .gitignore | uniq -d)
>    if [ "$duplicate_entries" ] ; then
>      echo "Error: Duplicate entries in .gitignore: " $duplicate_entries >&2
>      exit 1
>    fi

Thanks, that's right.
I've folded it in a more general form into the attached new
version of the patch, i.e. s/.gitignore/$file/g, together with
a "Improved-by" line:

+  duplicate_entries=$(sort $file | uniq -d)
+  if [ "$duplicate_entries" ] ; then
+    echo "Error: Duplicate entries in $file: " $duplicate_entries >&2
+    exit 1
+  fi

> BTW changes to bootstrap go through gnulib first
> and are synced from there, so when this is sorted
> (no pun intended) I'll propose it to gnulib too.

Thanks. A little merging will be necessary ... since the
introduction of warn+die in gnulib's bootstrap.

> p.s. While this removes the sorted requirement, it still
> sorts the file on insertion. That's not a new restriction
> but it does mess up/preclude the use of comments and separator
> lines in .gitignore. In future it might be better to
> auto generate something like this at the top of .gitignore:
> 
> ## Inserted by bootstrap ##
> entry1
> blank_line_after_last_entry
> 
> To do that, you'd also have to remove existing
> entries in .gitignore corresponding to the above section.

Yes, that sounds like a cleaner solution, although the details
may be tricky, e.g. if a manual entry contradicts to an
automatically added one. Can this happen?

Have a nice day,
Berny

gitignore-order-gcda-gcno4.patch
Description: Text Data