[lltx] [Fwd: Re: new tlmgr action]

[Manuel Pégourié-Gonnard]

Le 10/05/2010 13:10, Elie Roux a écrit :
> I think the main issue is timing here... let's discuss about it.
> 
Good idea, let's discuss.

> If I understand correctly, there is a version of luatexbase that is
> compatible with old versions of packages (it didn't work with me, but
> now that I have a version working, I don't really want to report bugs on
> the other), and there is also a new one introducing incompatible
> changes, like not writing in the callback nor tex table except for
> overriding callback.register.

Right.

> I totally agree with this changes. Where I
> disagree is when we should introduce those changes.

Well, since I never gave an opinion on that point, I fail to understand how you
can disagree with what I didn't say. So, just as I'm asking you to make me a
clear bug report and letting me reasonable amount of time (like 24h) to fiw it
before forking everything, I'm also asking you to ask me for my opinion before
disagreeing with it.

> If they are to be
> made (I think everyone agrees on it), why not changing it now, we have
> plenty of time to fix the other packages according to these changes.

I agree, but as I said in my other message minutes ago, the decision is not mine
to make, since the burden of adapting code in luaotfload, fontspec and other
packages relying on luatextra that I'm less aware of (like gregorio) isn't mine
either. So i'd like to have everyone's opinion on that.

I tend to agree with you: now would be a good time if it fits everyone's agenda.

> possible bugs (especially the huge one from
> http://github.com/eroux/luatexbase/commit/1b9ba1489eff995eedcc2c8bf41026cbedc75be9),

It happens only when both foobar.tex and foobar.lua exist. Also, your fix had
typos that made the code silently not work as advertised, and you didn't
document it. You also "fixed" a non-problem in the same commit.

Anyway, bugs happen and that's precisely why I asked everyone to bug me if
something stopped working when I released the fixed version. That doesn't make a
reason to fork after only a few hours of this bug not being fixed (while it has
been there for months).

> this gives me a bad feeling about pushing a version in TeXLive 2010 that
> noone tested except for you.

It has been tested by others, since I got reports about that. But of course I
agree that it's desirable to get as testing as possible. I took some time to
write a long message explaining things (2010603-28). If it wasn't clear enough,
I apologize for it, but you could ask me for clarification and wait a few hours
for me to answer. Btw, that's what happened, except that you though it was
appropriate to complicated the situation even more by forking because I promised
a clarification for today but decided to go sleeping first...

> Also, the reason I took the master branch
> is that in the master of luatextra, luamcallback dissapeared, but it is
> called by the compatible version of luatexbase, which means that the
> compatible version of luatexbase is in fact incompatible with the new
> version of luatextra... it certainly makes a lot of sense to you,

No, before your last commit, master of luatextra was compatible with luatexbase
tag 0.1 (now branch 0.1x) and luaextra's master as advertised.

> Summing up, I don't agree to work on several versions of all the
> packages at the same time, we should push all the (incompatible) changes
> now and test against the LuaTeX of TeXLive 2010.

I also think this is the way to go.

FYI (in case you want to try to understand and not only criticize) I was waiting
for Heiko's reply on luatex.sty and luatex-base merging, which could have
introduced more incompatible changes, before


> Otherwise you'll be
> alone to know the details of the compatibilities...

I have been trying hard to communicate on it, by email on this list and the
luatex list, and in the Changes file in luatexbase.

> But if I understand
> your last mail correctly, you're in the TeXLive team so you're the
> boss...

???!! I *never* said or *even thought* something like that! Insinuating that I
want to abuse my position as a TL developer is insulting! Will you please stop
insulting me? I'm trying to have a technical discussion here.

My point was, I'm strongly under the impression that you don't try to understand
my work and plans concerning luatexbase before modifying it. I was just asking
you to consider that if I made a change in the code or worked in some way, I may
have reasons for that, and these reasons may be valuable even if you wouldn't
have done things the same way, so please make the effort to discuss with me
before forking things and basically reverting my commits. This is common
courtesy, and I think I deserve it.

> I'm fine with it, I have a working version on my computer now, I
> can finally work.
> 
You could have achieved the same results with only a few hours difference, by
waiting for me to fix it as I told you yesterday I would do today. This would
have saved us some painful words too.

Manuel.