cvs commit: ports/archivers/hs-zip-archive Makefile distinfo pkg-descr pkg-plist ports/devel/hs-binary Makefile distinfo pkg-descr pkg-plist ports/devel/hs-darcs/files patch-Setup.lhs patch-darcs.cabal ports/devel/hs-darcs Makefile distinfo pkg-d

[Dmitry Marakasov]

* Gabor PALI (pgj at FreeBSD.org) wrote:

> > If there are executable files, that's an application regardless of to
> > what audience it's targeted (plain users or developers). If there
> > are no executable files, that's a module or library.
> 
> What if there is both?  Even worse: what if the executable is just an
> option (see archivers/hs-zip-archive) or it can installed as a library
> or as an executable or both (textproc/(hs-)pandoc)?  Most Haskell

I'd think of `primary purpose' here. If the application is optional,
that's most likely a module. Sometimes there may be applications
inrended only for use with module, they don't count too.  hs-zip-archive
seems to need a prefix, pandoc doesn't seem to. Basically, the
presence of prefixes was correct before the rename - those ports were
created by someone and followed common rules.

> Most Haskell applications has their own libraries which might be
> used by another applications (see textproc/(hs-)pandoc).

Most python, java, erlang apps do the same, so?

> > That guarantees nothing.
> > portupgrade -R ghc-* does.
> 
> I think no.  Sorry, I do not know portupgrade, so I have to branch my answer:
> 
> - If -R use build dependency for tracking updates: GHC is not updated
> as frequently (ab. twice a year) as its dependent ports, so this seems
> like exploiting a side-effect of portupgrade.
> 
> - If -R use run dependency for tracking updates:  It will not update
> standalone Haskell applications, because theoretically they should not
> depend on GHC (practically they depend at the moment but without an
> exact reason), since GHC is effectively required at compile-time.

It's the latter, build depends are not recorded. But why would one need
to update standalone apps? I thought the issue was that ports are
'tightly connected' - in that case they would run-depend on each other.

> > We've had these two audiences for years, along with many ports, many
> > languages and many prefixes, and it all worked well with established
> > rule of thumb for using or not using the prefix.
> 
> Oh, good to know.  Can you give me a pointer to this rule of thumb?

Ports collection. Just run pkg_info and look what is under py26- and
p5-. Then look for end-user apps written in python and p5.

-- 
Dmitry Marakasov   .   55B5 0596 FF1E 8D84 5F56  9510 D35A 80DD F9D2 F77D
amdmi3 at amdmi3.ru  ..:  jabber: amdmi3 at jabber.ru    http://www.amdmi3.ru