Re: [PATCH] gnulib-tool.py: Follow gnulib-tool changes, part 69.

[Collin Funk]

On 3/23/24 7:15 PM, Bruno Haible wrote:
> :-D Well, that's the principle of a von-Neumann computer: that a program
> (consisting of instructions) can assemble other instructions, to form a new
> program. For CPU instructions, it's a well-known technique. For 'sed'
> programs, apparently not...

LOL, that is a good way of looking at it.

> I haven't looked. I'm concentrating on the create-tests, and leave the
> import-tests to you.

Yes, that sounds like a good plan.

Sorry that this part is taking me so long...
I am starting to think that I have been looking in the wrong place the
entire time...

See this decade old typo:

diff --git a/pygnulib/GLImport.py b/pygnulib/GLImport.py
index d417137df6..2434e5f506 100644
--- a/pygnulib/GLImport.py
+++ b/pygnulib/GLImport.py
@@ -1393,7 +1393,7 @@ AC_DEFUN([%s_FILE_LIST], [\n''' % macro_prefix
             # Treat gnulib-comp.m4 like an added file, even if it already 
existed.
             filetable['added'] += [joinpath(m4base, 'gnulib-comp.m4')]
             filetable['added'] = sorted(set(filetable['added']))
-            filetable['removed'] = sorted(set(filetable['added']))
+            filetable['removed'] = sorted(set(filetable['removed']))

Then when creating the list of "dir|op|file" gnulib-tool.sh pipes it
into this:

LC_ALL=C sort -t'|' -k1,1 > "$tmp"/fileset-changes

but gnulib-tool.py doesn't do any sorting. Hopefully I am on the
correct track now atleast.

Collin