Subject: Re: [boost] [testing] Need optimization help!!!
From: Robert Ramey (ramey_at_[hidden])
Date: 2017-02-01 12:58:33
On 2/1/17 9:23 AM, Rene Rivera wrote:
> On Wed, Feb 1, 2017 at 11:14 AM, Robert Ramey <ramey_at_[hidden]> wrote:
> The 40 minutes includes *only* generating the reports that we see on the
> web site. By this point all the tests have run. Key though is that this is
> generation for *all* the tests for *all* libraries times all the testers on
> both master and develop branches. So we are talking about a large amount of
> data to process and generate.
OK - that's helpful. I use Library Status to generate tests for all the
combinations I test, release, debug, static, shared, debug, release for
a couple of compilers. The report generation phase is not nothing, but
it's insignificant next to building and running the tests themselves.
So it's ever been a problem. Of course I have no idea if it similar or
different in speed to the official boost one.
My motivation for creating library_status from Baeman's original program
was to be able to compare results across compilers and across
configurations which is very helpful to me in finding the cause of
problems in my library. So it clearly identifies build features such as
release vs debug, static vs shared, compiler vs compiler. etc. And it
doesn't truncate the error messages - which has caused me no end to
frustration on trying to use the boost report.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk