2016-10-15 14:39 GMT-03:00 Kevin Fenzi <ke...@scrye.com>:
> On Fri, 14 Oct 2016 17:33:00 +0100

>> If it is true .. maybe can someone help me to understand intention
>> packaging texlive like it is now?
> I can try and provide history here. ;)
> Disclaimer: This is just what I remember, it could be wrong.
> Texlive is unique. It's very handy and used, especially in the
> publishing and educational areas, so it's important to provide it in
> Fedora.

  There is a way to package texlive in small packages, using
few resources, etc. But you need to create like 3k packages
to bootstrap, the *really* hard part, then, create and retire like
2 to 8 per month.

> There's a number of similarities between texlive and perl's cpan or
> pythons pypy. It's a large collection of smaller packages that work
> with each other. However, it also has at least one big difference from
> those other collections: They do (about) yearly releases of the entire
> collection because it's very interrelated. Also, perl and python
> communities grew up around their packaging, so there's a large number
> of people packaging projects up as they are added. Texlive started as a
> free fork of tetex, so it appeared with tons and tons of packages and
> no community to package each little part up.

  Thats the big problem, or maybe an advantage if mapping rpm
meta packages to texlive meta packages.

> Long ago when texlive was replacing tetex, the first thing that had to
> happen was a legal review of all texlive. This took years and lots of
> people. Once that was done it got an exception to come in as one source
> package and lots of subpackages for all the reasons above. The orig
> maintainer had a program that generated the spec file, but they have
> long since moved on. Tom has taken over updating it recently and has
> done a lot to improve the spec. Yes, there have been bugs or issues,
> but sometimes thats the nature of things.
> So far this year:
> Added lines: 49387 Removed lines: 270572 Total # of lines: -221185
> So, Tom has removed 221k lines from there. I think thats a pretty good
> tally of improvement.
> So, focusing on positive actions here:
> * Perhaps we could split things up a little. I have often wondered if
>   just splitting things in 5-10 packages could help, but I would
>   absolutely defer to Tom here since he's been doing the work and he
>   was my sponsor 10 years ago and knows more about packaging than I
>   ever could.
> * If you have specific ideas for improvements, do file bugs and attach
>   your patches or explain what you think would help. I'm sure he would
>   love to hear it.
> Further than that I would wait for Tom to chime in...

  I talked a bit about at some point in the past, but I am currently
not very active contributing to openmandriva, where I did make a
1 to 1 rpm to texlive package, extraced %post scriptlets from
texlive installer, etc.


  The post script was a bit weird tough, it did run in background,
wait 15 secs, if some new texlive package was installed, wait
more 15 secs, and so on, and then, in background do the


  I even wrote a graphics texlive tool to install new packages or
update texlive:

that would use Mandriva urpm for actually managing the packages,
and look like this screenshot...

  The above might look like some self-promotion, maybe it is :)
but just to give an idea of a possible approach to handle it

devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Reply via email to