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:

> >> Like you suggested in your example on ports@ (but with "haskell"):
> >>
> >> # make -C /usr/ports search cat=cabal
> >
> > Why would a user need to specifically search for all cabal ports?
> 
> Because she wants to find darcs, for example.  If it is a hackage port
> it must be in that category independently of its name.

To find darcs, user should search for darcs, period.

> >> Is there any upgrade method that uses categories..?  A generic way
> >
> > Again, why would a user need it?
> 
> Because Cabal ports are just a subset of the "haskell" category.
> However, since the given port has been already installed, it does not
> really matter.  True.
>
> I think all the things boils down to the question of "where that port
> can be found".  Using a category might be overkill for this, but I do
> not see a better solution.
>
> But...  If you are right then some other ports (with "hs-" prefix at
> the moment) have "naming problems" like the ones I mentioned before:
> Agda, Agda-executable, alex, brainfuck, c2hs, cpphs, haddock, happy,
> hat, HaXml, hdoc, hmake, hoogle, hscolour, idoc, mueval, porte,
> texmath, unlambda, uuagc.

Probably, needs checking. Actually the other ports you've mentioned in
another mail (p5-Parde et al) should probably be fixed to.

> Actually, there is a STANDALONE variable in the Makefiles for really
> standalone Haskell applications (which means that the given port is
> not registered in the GHC package list), and the following ports use
> it at the moment: alex, happy, c2hs, hoogle, Agda-executable, pandoc
> (when no library installed), xmobar.
> 
> The not-so-nice thing is here when an end-user application depends on
> its "module" or "extension": devel/uuagc -> devel/hs-uualib or
> print/hs-hscolour-docs -> print/hscolour, devel/hs-haddock-docs ->
> devel/haddock, x11-wm/hs-xmonad-contrib -> x11-wm/xmonad.  Do you
> think we can do it without messing up things?

It can if you name it like that. I believe related ports should be
named with the same prefix. If hscolour, haddock, xmonad, then
definitely hscolour-docs, haddock-docs and xmonad-contrib. Though
probably OK for devel/hs-uualib.

> Let me guess about your answer: Rename all the connected ports as
> well.  But then it contradicts with the rule you mentioned.  Nice! :)

I never said these were rules - there can be no rules cause as you
can see they'll be conflicting. There's such thing as "common sence",
however. I believe that varies from person to person, but you should
try to target uses with as little knowledge as possible. You need
'darcs', but you don't even know which it is written in. Obviously
somecat/darcs. Well, I'm repeating myself.

I'll add that prefixes should be thought of as a mean to:
- resolve name conflicts between bindings to a single lib for many
  languages. I.e. this:

Port:   curl-7.20.0
Port:   curlpp-0.7.3_1
Port:   fpc-libcurl-2.2.4_1
Port:   ftp/ruby-curl
Port:   gambas2-gb-net-curl-2.15.2_3
Port:   hs-curl-1.3.5
Port:   linux-curl-7.13.1_2
Port:   linux-f10-curl-7.19.6
Port:   linux-f8-curl-7.18.2_2
Port:   ocaml-ocurl-0.5.3
Port:   p5-WWW-Curl-4.11_1
Port:   php4-curl-4.4.9_1
Port:   php5-curl-5.3.2_1
Port:   php52-curl-5.2.13_1
Port:   py26-curl-7.19.0_1
Port:   rexx-curl-1.4_3

- move ports that will only be needed as dependencies out of way for
  people who search for needed apps.
- ease it for developers to find libraries for languages they use,
  yes.

The priority of the last one is what we are arguing on - and I believe
we should tolerate towards plain users - they are likely a wider
audience, and they have less knowledge, thus it will be harder for them
to find hs-darcs, than for developers to find [nonhs-]somehaskelllib.

-- 
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