Re: commits for possible 5.9.1

>
> there have been 730 commits since 5.9 so there's a lot to choose from.
> are there any that we would specifically like to see in a 5.9.1?


Wow, over 700 yes/no decisions. That's huge. How should this happen? Do you
intend to look at each commit yourself? Or should we send you lists of
commits to include?

I could go through all my commits and let you know which ones should go in.
There was some work in 2022 about ERR_{EXIT,RETURN} that would be nice to
have. On the other hand, most of my more recent commits are about named
references. These should be left out (provided the whole feature can be
left out).


> The one obvious thing not to include is the nameref stuff.  (That's in
> any case not going to be perfect on its first release, so I don't think
> it needs to have every possible tweak before it appears --- it can be
> labelled experimental.  But if there remain obvious things to (re)implement
> or test there it can certainly wait.)  Hopefully it's not significantly
> affecting areas with other useful changes.
>

+1 for not including named references. In addition to the several pending
commits, I still have a long list of things that still need to be tested or
that I already know need to be fixed.

Another larger new feature are the no-fork command substitutions. Some
months ago, I started doing some exhaustive testing but then got distracted
by other things. From what I remember I didn't find any major issues, only
small things and maybe a few suggestions for doing things slightly
differently but I don't remember the details. In my opinion, no-fork
command substitutions could be included as an experimental new feature with
an indication that its design may still slightly change. This way more
adventurous users could play with it and let us know whether there are
things that could/should be added or done differently.

Actually, named references could also benefit from user feedback. There are
several aspects where multiple design options are possible. User feedback
could help us decide what should go into a final design for an official
(non-experimental) launch. So maybe include them after all? One thing that
may tip the balance to include them is if it's technically difficult to
exclude them because it causes many merge conflicts.

Philippe