[lsb-discuss] testing and more

Theodore Tso tytso at mit.edu
Sat Dec 23 15:52:22 PST 2006


On Sat, Dec 23, 2006 at 07:40:14AM -0500, Robert Schweikert wrote:
> The problem is getting the test into the right place, most ISVs will not
> know how to go about this and where or how to submit bugs for all the
> packages touched by the LSB.

Well, there is one easy place that ISV's can go to --- which is their
Linux Distribution provider, for which they've paid for a support
contract, presumably.  If you submit a test case which causes an X
server crash, presumably it will get fixed fairly quickly, and most
distributors are pretty good about forwarded the fix to the upstream
developers.

I could imagine a service (maybe offered by the FSG as a membership
benefit, or maybe offerred as a for-profit activity by some company)
which offered ISV's help in working with the upstream developer, on
the theory that this would get fixes and feature requests acted on a
bit faster than if you just contacted the distribution alone ---- but
somehow I can't imagine ISV's being willing to pay a lot for such a
service.  Maybe if it was coupled with "Linux development hotline"
(pay $X thousand dollars are year for 30 hours worth of "Linux
development consultant", plus $Y/hour for any help needed beyond
that), but presumably this would only be interesting to the
small-to-medium ISV companies --- the big companies can afford to hire
full-time Linux experts on staff who would be able to handle most of
these needs.  But maybe I'm all wet on this; what do you think?

> I would say these are example which should find their way to the LSB
> developer network site in some way shape of form. Maybe we need to have
> a "What not to do" section filled with examples showing how to break
> portability. Each would of course have to be commented appropriately.

Yeah, that would be a really great thing to have.  Unfortunately, some
of the best examples of what not to do are in propietary code and
trying to get companies to be willing to release their worst examples
of bad programming behavior is rather unlikely.  I'm in fact very
curious what in the world this unnamed application I mentioned earlier
had done to be so horribly tied to a single window manager, but since
we don't have source access, it's going to be very hard to determine
this.  

> Again the problem for most ISVs, who is the right person or the right
> list. IMHO we will have a hard time getting ISVs to expend the effort to
> dig many levels below the LSB.

Maybe --- how much though do you think the ISV's are going to be
willing to pay for this kind of service, though?  I'm not sure any
organization is going to be able to provide it for free....

(Well, you could submit a big report to the Debian Bug Tracking System
if you found the problem on Debian system, but there's no guarantee
they would respond in any kind of finite time.  But then again, you
get what you pay for; TNSTAAFL, you know.  :-)

						- Ted




More information about the lsb-discuss mailing list