Re: Doc package naming
Ian Lynagh wrote:
> On Wed, Jan 21, 2009 at 12:00:03PM -0600, John Goerzen wrote:
>> Ian Lynagh wrote:
>>> On Wed, Jan 21, 2009 at 11:04:11AM -0600, John Goerzen wrote:
>>>> For of all, I don't like libghc6-*-doc, because many of these packages
>>>> are useful not just for GHC, but also for Hugs.
>>> I was using libghc6-foo-doc because there were hugs libraries built from
>>> separate source packages. If they're unified then I'd use
>> That is true, but they aren't documented there, right? If you want to
>> see haddock docs for it, you've got to install libghc6-foo-doc. (And
>> why not; it would seem silly to have separate Hugs docs for the same
>> library.)
>
> In principle the docs could be different, e.g. hugs docs might not
> include parallelism-related functions. This leads to the deeper issue of
> packages not presenting uniform interfaces.
>
> Anyway, the right thing to do is just to build packages for all impls
> from teh same source package, and then using libhaskell-foo-doc is fine
> IMO.
>
>>> libhaskell-foo-doc.
>> Why not just foo-doc?
>
> Because I don't think we should be making packages called things like
> x11-doc
> readline-doc
> unix-doc
> time-doc
Right, but certainly in these cases the source package is haskell-x11
instead of x11, right? So haskell-x11-doc makes sense.
The "lib" prefix still doesn't quite make sense.
>
>>> Note that if ghc doesn't build on a platform then you don't have
>>> haddock, so can't update a doc index. You also can't create docs on such
>>> a platform. However, currently ghc builds on all platforms.
>>>
>>>> Secondly, why duplicate "haskell" in haskell-haskelldb-doc?
>>> Consistency.
>> With what?
>
> With the names used for packages which aren't called haskell* upstream.
But why no "lib" prefix then?
>
>
> Thanks
> Ian
>
>
Reply to: