Re: Help testing the pre-release of mc-4.8.18

[Andrey Gursky <andrey gursky e-mail ua>]

Hi,

On Thu, 29 Sep 2016 21:39:11 +0200 (CEST)
"Yury V. Zaytsev" <yury shurup com> wrote:

> On Tue, 27 Sep 2016, Paul Marwick wrote:
>
> > I can't say I was having any major problems with 4.8.17, though there 
> > was a second release from the Arch maintainer a few days ago with a 
> > couple of patches applied.
>
> I'm very grateful to the distro maintainers for adding patches to address 
> our screw-ups which have got uncovered after the release, but ideally 
> there should be simply less of those in my opinion.

I also appreciate the work of distro maintainers.

Regarding screw-ups. Yury, what do you think about issuing release
candidate versions of MC? This is a convenient way to get MC to be
tested in Distros on a regular basis (not git snapshots) (provided
maintainers update their packages in a timely manner). This is what,
e.g. Debian Experimental and Sid (unstable) are for, isn't it?

Additionally are these 4.8.x releases really only bug-fix releases? I
mean, if there would be at least one (and indeed there are some) commit
like refactor/rework/code-cleanup/... then it would be better to go
forward to 4.9.x and so on. The goal is to make .x really only small
bug-fix releases, while fixes, that going towards (even if only code)
improvements, aren't plain bug-fix releases anymore and should be
marked with a minor version increment, not a patch one. Then you
wouldn't need release candidates for true small/critical bug-fix
releases, though they are still possible.

> > I've built the package and have it running. So far, its working just 
> > fine. Are there any specific problems you would like checked? I use fish 
> > a lot (both within my own network and for remote work on customer 
> > sites), I use the editor and viewer and I make a great deal of use of 
> > the VFS support.
>
> Thank you for your feedback! This sounds good.
>
> I'm particularly worried about builds on platforms that we aren't 
> regularly using ourselves (non-Linux), like FreeBSD, Solaris, etc.

I'd like to propose to start/continue specifying where MC is supported.
I believe MC can be truly supported only on systems, that MC developers
are on, e.g. GNU/Linux (Debian, Fedora, OpenSUSE, Gentoo, Arch,..). Of
course this doesn't mean all others are out of care. I'd suggest to add
them into a "known to work on" / "not officially supported" section.
This means one doesn't have to worry to break something on such systems
anymore, just because MC cannot be tested there by MC developers (even
if there would be already existing tests). Once users of such system
find a bug, then this could be handled.

Saying, these systems are so important that we cannot risk to break
them (while having no access) seems contradictory IMHO. And it would
mean MC will be only cleanuped but no actual development will take
place?

> Anyways, so far it appears that there doesn't seem to be any serious 
> regressions, so we'll probably release 4.8.18 in the days to come...

Regards,
Andrey