[tex-live] pmx-2.6.19

[Robin Fairbairns]

Reinhard Kotucha <reinhard.kotucha at web.de> wrote:

> On 2013-01-15 at 17:01:34 +0100, Herbert Voss wrote:
> 
>  > Am 15.01.2013 16:53, schrieb Bob Tennent:
>  > >   >|> I know that and I wasn't asking for that.
>  > >   >|
>  > >   >|Then what are you asking for ?
>  > >
>  > > That a distribution like TeXLive that doesn't update executables
>  > > not install *inconsistent* upgrades. You realize that one doesn't
>  > > submit upgrades directly to TeXLive but to CTAN.
>  > 
>  > it should be no problem when you drop a line to the CTAN
>  > maintainers that they shouldn't trigger a CTAN-upload to TeXLive.
> 
> CTAN maintainers don't trigger TeX Live package updates.

quite so.  i was a little bemused by the thought that a message to _us_
would help.

> The update
> process at TeX Live isn't fully automatic.  It's still necessary to
> check the directory structure of packages on CTAN and to adapt the
> scripts accordingly.  Thus, packages are fetched from CTAN manually.

i don't know about "manually", but it's definitely under control by the
tex live team.

> In order to avoid such inconsistencies it's best to inform Karl and
> Norbert.
> 
> I must admit that I don't know much about MiKTeX.  But when I informed
> Christian about a new VnTeX release a few years ago, he told me that
> he fetched it from the TeX Live repository.  If he does the same with
> other packages too (anything else doesn't make sense, IMO), then
> Phil's suggestion (postpone until TL is frozen) is quite reasonable.

except that proceeding that way means that the update will never reach
tl.  if the update is postponed until tl is frozen, it's been postponed
to the point when nothing more can be added to tl.

unless it's possible to declare that the macro (etc) support for a
binary shouldn't be installed until the binary itself is installed, such
a package can never make it to tl.

perhaps the simple solution is never to upload to ctan.  which is a bit
sad, but if this is the way things work, it's inevitable.

i'm sure this stupid conclusion can be avoided, some way other than
never uploading the package (and, presumably, deleting the current
instance on the archive).

however, you wouldn't want a conclusion proposed by me, would you?

robin

(writing as _a_ ctan maintainer, probably not representing the team.)