[indiana-discuss] [Fwd: [desktop-discuss] SFE binary packages for OpenSolaris]

[Alvaro Lopez Ortega]

Bart Smaalders wrote, On 13/12/2007 18:21:
> Alvaro Lopez Ortega wrote:
>
>> So, my points were:
>>
>>  - Having source packages and binary packages relations is really
>>    useful. In fact, it is something that has proved to be helpful in
>>    other OSes and that many users miss.
>>
>>  - There wouldn't be any *real* need to change the consolidation
>>    system. It wouldn't be very efficient because of the large builds,
>>    but again, that would be much better than missing the feature. And,
>>    in fact, we are using LOT of F/OSS software that is not based on
>>    that sort of consolidations.
>>
>>  - I would have said that the state of the art kind of forced us to
>>    implement it. Even if you wanted to think of it like a
>>    "nice-to-have", it is a very important one if you think that we're
>>    targeting developers.
>>
> 
> You're missing the point.  ON developers don't build from source
> packages; they'll build from a mercurial workspace.  Why would
> anyone who's not working on ON compile it from source, and why
> should the OpenSolaris community spend the man years needed to
> support this?

It is quite clear to me, actually. We definitely should do it because it will 
allow people to develop on -and for- OpenSolaris at least as easily as they do 
on/for a Linux based system. Let's face it, that is the state of the art: 
people have been used to type plain commands for complex actions like:

  - Install all the needed dependencies to compile a certain package
  - Download a package source and rebuild it with these custom flags

Missing that kind of features would be simply a handicap for the developers of 
our platform. For instance, imagine that I want to fix a problem in 
Thunderbird; what would I have to do in order to get all the needed libraries 
and header files in place? and what about the sources, patches and files?

Think of this as the "familiarity problem" that Indiana is trying to address 
(the reason why IPS has been written in the first place), plus a missing 
functionality that impacts developers (who are an important target; aren't they?).

-- 
Greetings, alo.