[lsb-discuss] Linux Application Checker impressions
Wichmann, Mats D
mats.d.wichmann at intel.com
Sat Aug 9 11:45:10 PDT 2008
Hohndel, Dirk wrote:
> On 8/8/08 10:56 PM, "Wichmann, Mats D" <mats.d.wichmann at intel.com>
>>> 2) it was delivered as a tarball rather than an LSB package.
>>> While there's nothing wrong with tarballs, this
>>> seems like an area where eating your own dogfood
>>> might be appropriate.
>> It was discussed; while I generally agree with you this
>> was a case where (for now) it was considered it could
>> provide a barrier to entry for people who don't otherwise
>> have an LSB-conforming distro to use for their testing.
> I think that's the wrong assumption. LSB is everywhere - there are
> basically no distributions out there that aren't either LSB out of
> the box or can be made LSB conforming by installing a couple of
> packages. So it makes only sense to have this an LSB package.
> And in a way this is a matter of "attitude" as well. Delivering it as
> tar ball implies that we don't believe in our own success...
While personally agreeing with you, I'll make two observations
that weighed in on this, plus add that I believe this delivery
mechanism is intended to be temporary.
1. we've been told there are developers who aren't allowed
root access to their boxes to do package installations, and
must go begging to their IT people to get anything like that
to happen; thus a desire to have a mechanism where you can
just unpack files in your directory and use them.
2. while distros are able to provide the LSB environment,
if they don't, the failure mode of LSB binaries is very
complete and very non-intuitive, something like:
bash: No such file or directory
Referring, in fact, to the absence of the linker, not
the program binary. I notice recent Fedora is now giving
a much better error message:
bash: labappchk: /lib64/ld-lsb-x86-64.so.3: bad ELF interpreter:
No such file or directory
If that was universally the message this point could go away.
More information about the lsb-discuss