Re: pkg/48919

[David Sainty <dave%dtsp.co.nz@localhost>]

The following reply was made to PR pkg/48919; it has been noted by GNATS.

From: David Sainty <dave%dtsp.co.nz@localhost>
To: Philip Miller <sighoya%gmail.com@localhost>
Cc: gnats-bugs%NetBSD.org@localhost
Subject: Re: pkg/48919
Date: Mon, 07 Jul 2014 18:42:24 +1200

 On 06/07/14 02:04, Philip Miller wrote:
 > On 07/04/2014 01:06 AM, David Sainty wrote:
 >> It looks right.  Are you able to check whether the installed 
 >> freetype2.pc also looks like this?
 > Is the same.
 >>
 >> And I guess then try clean+build in fontconfig - I know you've done 
 >> this, but I'm just thinking accidental stale build state might be 
 >> tripping us up.
 >>
 > cd fonts/fontconfig && bmake clean build
 >> If fontconfig clean+build still fails for harfbuzz reasons, perhaps 
 >> the config.log from fontconfig might help (should be output somewhere 
 >> in the fontconfig work directory)... 
 > Cause of unknown reasons, the build passed. I have made exactly the 
 > same before, I don't understand...
 > I have tried to install more packages, which require freetype2, and 
 > they passed the build.
 > So I think the problem is solved.
 > Thanks a lot David 
 
 Excellent.
 
 Ok, so the problems were:
 
 1. native freetype2 with built-in png support is not supported by Pkgsrc 
 at the moment.  This is not ideal, and presumably is going to be more of 
 a problem over time.  No good solution for this yet.
 
 One solution might require native support for png, but wiz@ has sensibly 
 expressed displeasure at this.
 
 Manual workaround is:
 
      PREFER.freetype2=pkgsrc
      PREFER.fontconfig=pkgsrc
 
 2. With the above workaround, fonttype2 fails to build if there is a 
 native harfbuzz on the system.  The solution for that is to explicitly 
 disable harfbuzz support in fonttype2.  I'll fix that in Pkgsrc.
 
 3. With that fix in place, you were being bitten by stale build state.  
 That's easy to do when making changes to packages, but it's very much a 
 one-off.  We can ignore that problem as it's purely a consequence of 
 debugging the problem.