cvs commit: ports/archivers/hs-zip-archive Makefile distinfo
pkg-descr pkg-plist ports/devel/hs-binary Makefile distinfo pkg-descr
ports/devel/hs-darcs/files patch-Setup.lhs patch-darcs.cabal
ports/devel/hs-darcs Makefile distinfo pkg-d
pgj at FreeBSD.org
Wed Jun 2 16:11:30 UTC 2010
On 06/02/10 17:31, Dmitry Marakasov wrote:
> Hm, I feel double about this. On the one hand, it's nicely splitting
> library from implementation, on the other hand, it's overcomplicated (as
> most master/slave ports), just look at the plist.
It has a very similar plist to pandoc's... Not accidentally, because
that gave me the idea. Why is it complicated? I will only need to
maintain the master port, the slave is just a front-end. Where is the
complication here? (If you want to refer to the Makefile and pkg-plist:
They are just implementation details, are not they? -- Anyway, I think
it is possible that pkg-plist files can be replaced by autoplists 
similar to the Ruby guys' solution.)
> Also, will both ports have the same option list? If yes, what happens if you specify
> conflicting options for them?
Yes, they have the same option list but with different stored
configurations (thanks to their different names). When you specify
conflicting options for them, then they will be built with two different
configurations without any problems.
> Building it twice also doesn't feel good.
Your average non-Haskell-aware user will not build the port twice, you
can take it for sure ;)
> It'd probably be ok if it was two separate ports, building only needed
> stuff and having separate set of options. However, I still think that
> the best would be to just have "darcs" that "installs modules as a
Here hs-darcs installs darcs as bonus. What is wrong with this? Again,
your average non-Haskell-aware user does not want to see those modules
(but perhaps some other ports...), and darcs works fine without them.
> like some other end-user applications do (and some even install
> modules for several language at once).
For example? I need specific examples to slurp solutions from them.
Sorry, but your previous vague references have not proven to be so valuable.
> The only drawback is not having hs- prefix, which as I told, is not expected by users, and should
> not be expected even by haskell programmers.
Have not you read the mail where I reasoned about the split ?
More information about the cvs-ports