[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: cduce_0.3.2-1_powerpc.changes REJECTED



On 10258 March 1977, Thomas Petazzoni wrote:

> As far as I understand it, it appears to be that the rpath problem (as 
> explained in http://wiki.debian.net/index.cgi?RpathIssue) is not an 
> issue with Ocaml. In fact, the ld.so and gcc things are now involved 
> with these rpath, it's up to the Ocaml virtual machine to handle them.

If it points to /usr/lib/package/ then yes. usr/local is only for local
admins.

On 10258 March 1977, Stefano Zacchiroli wrote:
>> >needed, but usr/local/lib? No.
>> As I'm a new packager, and very new to Ocaml stuff, I've asked 
>> experienced Ocaml packagers about this issue. It appears to be that it 
>> is a well-known issue, that has already been discussed on the 
>> debian-mentors list [1] as well as on the specific debian-ocaml-maint 
>> list [2].
> Indeed, Joerg is right.

Thanks :)

> Ocaml packages have often rpath to
> /usr/lib/ocaml/[OCAMLVERSION]/stublibs since there are located .so
> objects dynamically loaded by the interpreter. This is ok since ocaml
> debian packaging ensures that these shared objects are located in that
> directory on every debian installation (same assumption that other
> packages do for supporting .so pluging installed on /usr/lib/SOMEWHERE).

Ive seen many of them, yes. People keep uploading new ocaml stuff. :)

> But /usr/local/lib is really strange and should be avoided, probably it
> is there for some upstream issues in the compilation phase.

At least it isnt obvious in the configure/Makefile thing (maybe only if
you actually understand this ocaml stuff). :)

> Just out of curiosity, why do you consider this "bug" important enough
> to stop a package enter the archive? I don't think it is that critical
> after all.

Well. Somewhere you need to draw the line[1]. :)
I could have just mailed the packager with the thing I have, so he can
upload a fixed one. Yes. But this way, with the reject, leaves him free
to upload it with a -1 again (together with source of course), ie no -2
changes[2] needed for it, as it just enters the archive then. Some people
may prefer this (and it actually happened this way with another package,
rejected two times but still went with -1 in the archive).

And: I *personally* dont see a reject as a bad thing, its not ultimate and
forever. If one disagrees: Answer to the mail with a reason and upload
again. We are all humans, we can actually change our mind. :)


[1] Hehe, if I would use the same ruleset I used as AM and sponsor, I
    could have rejected around 80% or 90% of the packages. :)
[2] If it is distributed to other people already, like your own website
    or so, its better to enhance the version, so they have a good
    upgrade path to the thing in the archive.

-- 
bye Joerg
Five exclamation marks, the sure sign of an insane mind.
			-- Terry Pratchett, Reaper Man



Reply to: