Re: [Caml-list] GODI is shutting down

[Adrien Nader <adrien@notk.org>]

Hi,

On Mon, Jul 22, 2013, Gerd Stolpmann wrote:
> Dear subscriber,
> 
> Unfortunately, it is no longer possible for me to run the GODI
> distribution. GODI will not upgrade to OCaml 4.01 once it is out, and it
> will shut down the public service in the course of September 2013. The
> website, camlcity.org, will remain up, but with reduced content. Existing
> GODI installations can be continued to be used, but upgrades or bugfixes
> will not be available when GODI is off.
> 
> Although there are still a lot of GODI users, it is unavoidable to shut
> GODI down due to lack of supporters, especially package developers. I was
> more or less alone in the past months, and my time contingent will not
> allow it to do the upgrade to OCaml 4.01 alone (when it is released).

It is very sad to hear that. I was expecting some clashes for the past
few months so I'm not completely surprised.
I've been meaning to do more for godi but I've been completely unable to
do so unfortunately and I'm sorry for that.

> Also, there was a lot of noise about a competing packaging system for
> OCaml in the past weeks: OPAM. Apparently, it got a lot of attention both
> from individuals and from organizations. As I see it, the OCaml community
> is too small to support two systems, and so in some sense GODI is
> displaced by OPAM.

I believe that this is the same as with LLVM and GCC. Two years ago,
llvm was supposed to beat GCC within a year: performance of the
generated code was increasing very quickly.
It was a new software, not hampered by decades of development, and the
rule of 80/20 struck. It was fairly obvious that the rate of
improvements would slow down and that people would start asking for more
stability and compatibility (to this day, llvm still doesn't have minor
releases).

It must have been hell for GCC people back then. Noone cared that they
were more compatible (for hosts, targets, options and combinations of
these), more stable, generated faster code, had been around for decades,
...

> The sad part is that OPAM is only clearly better in one point, namely in
> interacting with the community (via Github). In times where social
> networks are worth billions this is probably the striking point. It
> doesn't matter that OPAM lacks core functions like deleting all files when
> a package is removed, and that it lacks many other features GODI has. So
> there is some loss of functionality for the community (partly difficult to
> replace, like GODI's support for Windows).

To be honest, I've never understood why opam was "started". I believe
the time would have been much better spent improving godi.
I've heard complaints against the UI and the package release process of
godi.
How can such complaints about the UI be a reason to start a new
project?! Couldn't the package release process be improved gradually?

As for github, I concur its best feature is the "social" thing. And
sometimes it gets in the way: you can see people who've forked your
repo and their commits easily. But for other things, cgit is way better
and clearer. Sure, it doesn't have a shiny CSS. On the other hand, it
doesn't put flash objects in webpages like github.

> If somebody wants to take over GODI, please do so. The source code is
> still available as well as the package directories. Maybe it is sufficient
> to move the repository to a public place and to redesign the package
> release process to give GODI a restart.

You haven't only written the code but you've also setup a whole
infrastructure. Would you mind trying to pass it over or migrate it,
maybe not leaving completely but letting others do some of the annoying
admin-related tasks?

> There is also another point that was driving me mad in the past weeks,
> namely missing respect from the OPAM guys. Given the fact that OPAM is
> only a thin layer around ocamlfind (and guess who wrote it), and given the
> fact that GODI was pioneering in many fields, I was expecting nicer
> wordings, and less dumb campaigning ("we have 400 packages, and you only
> 170"). OPAM is only harvesting what I seeded many years ago.

I tend to agree here. Something like 5 years ago, management of OCaml
library was a real pain. Ocamlfind support was everything but universal
(I remember when ocamlfind-enabled build systems where the minority and
I found them annoying). I believe godi really helped with that but it
also had to support build systems not using ocamlfind and that made some
godi packages more complicated.

The people involved in OPAM have clearly been more vocal and I'm
definitely not going to blame them for that: if you build a new project
but don't advertise it, noone will use it. Some of the messages were
probably lacking some tact however.


As far as I'm concerned, I don't want to use OPAM. I've been happy with
GODI and would really like to continue to use it. I do not want to use a
high-speed source-based rolling-release system and I believe that it is
bad, especially for beginners who need a stable environment.

I've been very irritated by pieces of advice for beginners that
mentionned using svn branches of the OCaml compiler. OPAM makes it easy;
great! It's still an unreleased version that is still in development.
Generally speaking, I find the use of bleeding-edge software bad. It
might be good OCaml libraries that are well-tested but it doesn't
matter: you're going to get into troubles sooner or later.

I'm not advocating waiting for years for things to settle but vim
$package, :%s/0.14/0.15/g, :wq, git commit -a -m 'foo: Update to 0.15.',
git push might be too quick.

I also believe "pinning" is a bad idea in general: it picks bits and
pieces of software from different timeframes. It will work for
exceptional cases but otherwise you end up with an exponential explosion
of combinations of versions that haven't been tested together.

There could be a git branch or a git repo for stable stuff along with
the development tree. Citrix already does that and others already do
that too. Which means that there are several parallel distributions and
that has its drawbacks too.

Obviously, all of these can be fixed or improved, through code, process
and probably better: a combination of both. However, currently, there
are still open issues.

I've lacked time recently and I'm still using godi with 3.12.1. And I
can still update it and it works as well as the 4.00 branch. I doubt
that I could get the same with OPAM (had it started with 3.12).

-- 
Adrien Nader