[gobolinux-users] Source repositories and other suggestions

Martin Baldan martinobal at gmail.com
Sun Dec 3 12:51:56 UTC 2006


On 12/3/06, Michael Homer <gobo-users-dufus at wotfun.com> wrote:
>
>
> Yeah, I was thinking that that's how they'd actually be transmitted.
> It makes for a complicated search though, which is why I thought using
> a tree would be easier. I guess it isn't actually necessary to do
> anything beyond a yes/no result for any one set.


That's the idea. A compatibility table would have as keys the R-R name of
each application or library, and as values it can have a few fields of
information about tests (numbrer of possitive tests, number of negative
tests, approved by a dev?, ...). I don't think you need anything else, but
of course you can build a tree from that if needed.


>
> What I was getting at is that people will just accept the test that
> came with the recipe, and not bother contributing more until a new
> version comes out.


Well, that's  probably true, but it very much depends on the effort
required. What I have in mind is that, when an app is still very new (few
tests), every time a user closes the app, a prompt or a pop-up appears,
asking "did this app work correctly (yes/no)", so it only takes one click to
answer. When a few of these tests are done, the pop-ups stop appearing to
anyone. Of course, users who want it should be able to refuse to cooperate
and get no pop-ups ever.


> I'm not sure it would be that difficult to do a simple version of it.
> It boils down to recursively running ldd and creating a tree. I guess
> parsing out the dependencies and then building the hypothetical
> what-if-I-install-X would be a bit more complicated though. If you
> know Ruby, looking at some of the code in Freshen might help, or in
> Manager, although neither of them addresses the bulk of what you'd
> need here.


Ok, thanks for the tips ;)

Is this going to require some sort of local database of what
> everything is compiled against? That sort of goes against the
> filesystem-as-package manager philosophy too.
> _


Well, yes, you have the info about sonames-TO-real_names in the filesystem,
but you need a table of real_names-TO -RR_names, so that you can look up the
R-R names in the compatibility table.
At first, the real_names-TO-RR_names table would have null values, and it
would start replacing them with real values as you compile applications. Or
you could tell the system to recompile every program, and then have a valid
table.

So, it's not that bad, but yes, I have the same kind of feeling. A more gobo
way would be something like sonames point to real names point to R-R names,
but this seems to be too disruptive. Maybe there a way to do it. The good
news is that this is independent from the compatibility table, so we can
rethink it later.

I've just seen a suggestion from Jonatan Lilijedahl. It looks interesting,
but I don't know its exact meaning an implications.

--Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.gobolinux.org/pipermail/gobolinux-users/attachments/20061203/9bf5722d/attachment.htm 


More information about the gobolinux-users mailing list