[oe] [meta-oe/meta-webserver][PATCH v4 00/13] Merge modphp into php and upgrade

Paul Eggleton paul.eggleton at linux.intel.com
Fri Jan 9 10:22:24 UTC 2015


On Friday 09 January 2015 10:38:10 Martin Jansa wrote:
> On Wed, Jan 07, 2015 at 02:50:15PM +0000, Paul Eggleton wrote:
> > Finally resolve the overlap between the modphp and php recipes by
> > merging the former into the latter. Along the way I fixed a few minor
> > issues and upgraded related recipes.
> > 
> > I've taken a stepped approach to doing the php merge / upgrade so it's
> > easier to see what's going on and easier for me to test - first upgrade
> > php to the same version as modphp, then do the integration, then upgrade
> > to the latest 5.5 release. I would appreciate someone testing PHP in CGI
> > mode however since I don't have a convenient test setup for that (as I
> > do for mod_php).
> > 
> > Changes since v2/3:
> > * Fix phar script installation and packaging, which also resolves a QA
> >   warning
> 
> There is one more issue:
> 
> sstate-sysroot-crutf.sh script was whitelisting this entries:
> [^/]*/usr/lib/php/\.channels.*
> [^/]*/usr/lib/php/\.registry.*
> [^/]*/usr/lib/php/\.depdb.*
> [^/]*/usr/lib/php/\.filemap
> [^/]*/usr/lib/php/\.lock
> 
> No it's in php5 directory so it's easy to change the whitelist, can you
> please confirm that it's safe to ignore these files (that nobody will
> miss them if they aren't there after sstate reuse) or check if we can
> prevent creating them in first place?

I'll have to look into this, I know next to nothing about PEAR at this point. 
My gut feeling is that they could still be whitelisted without ill effect; 
perhaps they could just not be installed at all.

> And is "/usr/lib/php5/php" expected, shouldn't it be just
> "/usr/lib/php5"?

I agree this looks odd, but it seems it's how it ends up with libdir set in 
order to make the extensions directory be more sensible than the default 
/usr/lib/extensions (which is a worse alternative than the current state IMO). 
I couldn't immediately see how to fix this when I looked at it earlier.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre



More information about the Openembedded-devel mailing list