Hi,
On 2015-12-17 01:50, Frank Ch. Eigler wrote:
> kenj wrote:
>
>> [...]
>> You're already in unchartered water here ... I've never tested the
>> --prefix=... option for a build and would not be surprised if this has
>> problems.
>>
>> I don't think this should be necessary, nor should it be attempted
>> ... the PCP QA suite is designed with a philosophy that it is trying
>> to exercise the code in a context that is as close as possible to that
>> which an end-user would experience. [...]
thanks for the clarification added recently to the qa/README file, it
now sets expectations much more clearly than before. I think it could
perhaps still be clarified a bit that running QA also in installed but
not packaged location is unsupported (in addition to running QA in
source tree) but already the current README version is ok.
> For the record, --prefix is not incompatible with end-user experience.
> We added --prefix to the configury specifically because there was a
> use-case for installing and running pcp out of a non-core-os
> subdirectory. And it worked fine, including /etc/rc.d files that
> redirected to the $prefix/bin programs. I believe we ran pcpqa that
> way too.
>
> There were only a few configuration prereqs like putting $prefix/bin
> into the $PATH (ahead of any conflicting system copy). For
> pcp-libs-devel purposes, an end-user $CFLAGS would have to include
> -I$prefix/include, etc. - something we could automate with
> http://oss.sgi.com/bugzilla/show_bug.cgi?id=1095 .
>
> Embracing --prefix more could also pave the way for letting pcpqa run
> on normal workstations, with much less disruption to the system, thus
> making it more pleasant to be run frequently.
Agreed, the need for packaging and a dedicated test VM is considerable
overhead when comparing running the test suites of other projects. I'm
not sure are there any fundamental reasons which would prevent running
at least parts of the PCP QA suite without packaging. I understand that
there's a lots of test cases where a proper installation is a must to
have meaningful results but there are also quite a few cases that should
not be needed.
Thanks,
--
Marko Myllynen
|