Subject: Re: [boost] Windows Develop Tests Failing
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2015-03-13 20:30:24
On 13 Mar 2015 at 16:18, Steven Watanabe wrote:
> > (c) As an alternative, bjam could have an option to compress the path to a
> > shorter hash value.
> b2 --hash (reduces the path to an md5 hash)
> b2 --abbreviate-paths (shorter but still somewhat human readable)
This won't help when b2 is itself called from within a too long path.
The Jenkins matrix project plugin is especially bad for this.
To workaround that problem, have Jenkins create a junction point
close to the root using mklink /j. Change directory into the junction
point, and work from there.
The only big downside to that technique is you must serialise jobs on
Windows slaves, because two jobs will try to set the same junction
point and it races. I solve that by dropping the executors to 1, and
firing up more than one Windows slave VM. Kernel memory sharing
reduces the RAM cost of this very considerably, so it's not a
-- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk