Re: Trusted maintainer policy notes (draft for discussion) (was Re: [ATTN maintainer] libunistring 1.3 needed)

On 01/02/2026 14:20, Jon Turney via Cygwin-apps wrote:
> On 24/01/2026 20:44, Brian Inglis via Cygwin-apps wrote:
>> On 2026-01-24 11:10, Marco Atzeri via Cygwin-apps wrote:
[...]
>>>
>>> it looks fine for me.
>>>
>>> Only question about Alphabetic order, which one ?
>>> Today it seems an almost ASCII order where Uppercase and Lowercase
>>> are all considered Uppercase.
>>>
>>> If I do a "sort" (LC_CTYPE=it_IT.UTF-8) and compare
>>>
>>> @@@ -4136,8 +4137,8 @@
>>>   xtrans                                       @xorg
>>>   xvidcore                                     Takashi Yano
>>>   xwd                                          @xorg
>>> -xwin-xdg-menu                                Jon Turney
>>>   xwininfo                                     @xorg
>>> +xwin-xdg-menu                                Jon Turney
>>>   xwud                                         @xorg
>>>   xxdiff                                       ORPHANED (Yaakov 
>>> Selkowitz)
>>>   xxhash                                       Marco Atzeri
>>>
>>> the position of "-" and "_" seems different depending on the Local used
>> That seems to be similar to the order in setup.ini - strcasecmp or 
>> rpmvercmp?
>   Yes, this seems to be an attempt at the same ordering as used in 
> setup.ini. [1]
> 
> [1] https://cygwin.com/cgit/cygwin-apps/calm/tree/calm/package.py#n658
> 
> I guess ignoring case is maybe kind of helpful, so you don't have to 
> remember the correct casing to locate an entry,
> 
> But yeah, the point that if I'm going to ask people to keep the list 
> sorted, what the correct ordering is exactly should be well-defined, is 
> excellently made :)

I've put an initial attempt at a hook in place, which does:

git show $newrev:cygwin-pkg-maint | sed '/^[@#]/d' | LC_ALL=C sort -f 
-k1 -c || exit 1

... that is:

discard all lines starting with @ or #
check that the file is sorted on the first whitespace-delimited field, 
case-folded.

... which seems to be something close to what we currently have.


Please let me know if there are any problems with that.