Re: guile-2.0 and debian

[Antonio Ospite]

On Wed, 23 Nov 2016 09:34:05 +0100
David Kastrup <address@hidden> wrote:

> Antonio Ospite <address@hidden> writes:
> 
> > On Wed, 23 Nov 2016 00:25:03 +0100
> > Thomas Morley <address@hidden> wrote:
> >
> > [...]
> >> Hi Antonio,
> >> 
> >> I figured to do a regtest-comparison between builds with guile 1.8.8
> >> and guile 2.0.13:
> >> 
> >> For that I had to get back guile 1.8.8 and did a build from current master,
> >> then I did 'make test-baseline'.
> >> Then I copied the entire folder 'lilypond-git/build/input' elsewhere.
> >> 
> >> As second step I got guile 2.0.13 back
> >> (Which is pretty tedious, because it's not in the distro, even not for
> >> Ubuntu 16.10, if I'm not mistaken.)
> 
> Isn't 2.0.12 sufficient?
>

It should be, yes, guile-2.0.13 fixed some security issues but I didn't
see bug fixes relevant to lilypond.

However as others have said Debian never packaged 2.0.12 and it is going
to have 2.0.13 in the next stable release so 2.0.13 is the most
convenient to use, at least for me and other debian users.

But I agree that the autoconf check can be >= 2.0.12.

[...]
> > You could install debian stable in a virtual machine.
> >
> > Or for a more lightweight approach you can create a debian stable tree
> > using debootstrap and run a shell from it in a container with
> > systemd-nspawn, this is what I did for my quick tests with guile-1.8.
> >
> > The same goes for people wanting to try lilypond with guile-2.0.13, in
> > that case a debian unstable container is to be used.
> >
> > I can elaborate more if there is interest.
> 
> The question is whether it would make sense to temporarily base lilydev
> on something with the necessary packages instead of vanilla Ubuntu.
> There is a bit of impetus for getting a hold of the Guile-2.0 issue and
> I find that expanding the base of people willing to dig into matters
> would be a useful thing.  It might also improve chances of getting
> actual Guile developers touching our problem spaces.
>
> Having the kind of work Thomas invests here be doable with straight
> lilydev could draw some more participation.
>

Ah I didn't know about lilydev (https://github.com/fedelibre/LilyDev).

Updating it to Debian testing aka Stretch (the _next_ Debian stable
release) will expose people to guile-2.0.13.

Federico, AFAICT the current 4.1 works fine for guile-1.8 (FTR Debian
stable has both 1.8.8 and 2.0.11, but lilydev only provides the former),
maybe if you release a version based on Stretch use 5.0 as version
number to make it easier to differentiate between the two.

BTW for those who don't want to run a full virtual machine, a
lightweight lilydev container can be created and launched with the
attached script.

> And it's very likely to be an area of the "the last 10% take 90% of
> fiddling" kind where "it almost works" is a good incentive for further
> diggers.

I was saving the joke "80% done, 80% to go" for later, but you
anticipated me :) which is good as it means that we are on the same
page.

Up to now I hacked my way around the problems in my spare time just to
make lilypond build and run with guile-2.0 so that debian could package
it, but I do realize that my patches are far from being "mergeable".

If there any chances about having some paid time sponsored to fix things
properly I may consider doing it, feel free to contact me off-list.

Ciao ciao,
   Antonio

-- 
Antonio Ospite
https://ao2.it
https://twitter.com/ao2it

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?

lilydev-container.sh
Description: Text Data