[Devel] Thoughts on lv2-universe-YYYYMMDD.tar.bz2

[Ricardo Wurmus]

On 8 February 2012 15:16, Brendan Jones <brendan.jones.it at gmail.com> wrote:
> On 02/08/2012 04:13 AM, David Robillard wrote:
>>
>> I have wrestled with this suggestion for some time; whether to release
>> core and all extensions and examples etc. (everything in the repo,
>> essentially) in one tarball.
>>
>> Feedback is not so great from people who actually have to install all
>> this stuff, understandably.  I very much dislike the centralized feeling
>> nature of it, but there are arguments for:
>>
>>  * The reality of things is that every time I tweak the documentation
>> tools or whatever *all* extensions change
>>
>>  * Building and installing HTML documentation with cross-linking would
>> be possible.  With separate releases it is not.  Not installing
>> documentation is not so great.
>>
>>  * One build process to keep up to date from source (user, or developer
>> just dropping the whole thing in their tree) vs a looot
>>
>> The main against is I guess the lack of ability for fine-grained
>> installation, but having an extension installed really shouldn't cause
>> any harm in any case.  Separate releases make package dependencies more
>> informative as well.
>>
>> Any thoughts on this?  I get the feeling jumping on board LV2 is too
>> "hard", which suggest I'm not doing such a great job as maintainer.
>> Maybe a big ol' package of up-to-date everything would help this.
>>
>> -dr
>>
>> _______________________________________________
>> Devel mailing list
>> Devel at lists.lv2plug.in
>> http://lists.lv2plug.in/listinfo.cgi/devel-lv2plug.in
>
> It would certainly expedite any packaging in distros. That could only help
> adoption I would think.
>
> Brendan

I second Brendan's comment. In Arch Linux lv2core is available via the
official repository, but a number of extensions are not; they are only
available in the unofficial user repositories where one often finds
outdated versions. Keeping this all in one tarball guarantees
compatibility and would certainly be a good idea.

(Incidentally, the situation with Haskell looked similar before the
introduction of the Haskell Platform.)