On Thu, Sep 19, 2013 at 01:59:33AM +0200, Juan Francisco Cantero Hurtado wrote:
> On Mon, Sep 09, 2013 at 11:16:47PM +0200, Juan Francisco Cantero Hurtado 
> wrote:
> > On Mon, Sep 09, 2013 at 10:05:18PM +0200, Landry Breuil wrote:
> > > On Mon, Sep 09, 2013 at 08:11:08PM +0200, Juan Francisco Cantero Hurtado 
> > > wrote:
> > > > On Mon, Sep 09, 2013 at 07:17:33PM +0200, Landry Breuil wrote:
> > > > > On Mon, Sep 09, 2013 at 06:24:22PM +0200, Juan Francisco Cantero 
> > > > > Hurtado wrote:
> > > > > > On Mon, Sep 09, 2013 at 05:24:53PM +0200, Landry Breuil wrote:
> > > > > > > On Mon, Sep 09, 2013 at 04:03:43PM +0200, Juan Francisco Cantero 
> > > > > > > Hurtado wrote:
> > > > > > > > On Mon, Sep 09, 2013 at 10:11:26AM +0300, Lars Engblom wrote:
> > > > > > > > > I have been testing this on amd64 after time_t switch. It 
> > > > > > > > > works
> > > > > > > > > well. I tested even some date functions in racket/date.
> > > > > > > > > 
> > > > > > > > > Is it ready to be included in ports?
> > > > > > > > 
> > > > > > > > Racket has been ready for months but summer isn't the best 
> > > > > > > > season in the
> > > > > > > > open source world.
> > > > > > > 
> > > > > > > PFRAG.i386 and PFRAG.amd64 make no sense, there are lots of 
> > > > > > > common files
> > > > > > > there, and nothing seems arch-dependent.
> > > > > > > 
> > > > > > 
> > > > > > It makes sense. The files in PFRAG.* are generated from the source 
> > > > > > of
> > > > > > the scribble docs (also the png files). There is a bug in the 
> > > > > > source and
> > > > > > racket doesn't generate all files on i386, the file list is 
> > > > > > different
> > > > > > between i386 and amd64. The files are named with consecutive 
> > > > > > numbers, so
> > > > > > the content is different in some files despite of the same filename.
> > > > > > 
> > > > > > The easiest workaround is to trim the directory from PLIST and to 
> > > > > > use
> > > > > > PFRAG.*. I reported the bug to the doc author in February but he
> > > > > > never replied my mail.
> > > > > 
> > > > > All the common parts from PFRAG.* should go to PLIST to actually make
> > > > > sense. Noone cares about the content/cksum at that step.. since there
> > > > > are more files in PFRAG.amd64, PFRAG.i386 should be merged to PLIST, 
> > > > > and
> > > > > PFRAG.amd64 should only contain the list of images thare are only
> > > > > present there. But still, this is a gross use of PFRAG imo.
> > > > > 
> > > > 
> > > > I know my use of PFRAG is gross, I'm not proud of this :P . Thanks for
> > > > the help.
> > > > 
> > > > I've followed your advices and now the port only uses a minimal
> > > > PFRAG.amd64. Attached the tarball with the changes (untested).
> > > 
> > > That looks a bit better :)
> > > 
> > > lib/libracket3m.so handling seems wrong, it should be properly
> > > versionned within the ports infrastructure (if things are supposed to be
> > > linked with it) and there shouldnt be a -5.3.6.so file (symlink ?)
> > > Why isnt there an @lib annotation for it ?
> > > 
> > 
> > I thought the lib was only for internal use but I was wrong. I'm reading
> > the docs and it's possible to embed racket in other applications.
> > 
> > I'll fix the problem this week.
> > 
> 
> I fixed the problem with SHARED_LIBS. Let me know if you see more
> errors.

Reads good to me, if needed it can be tweaked in-tree. If anyone wants
to import that i'm ok with it.

Landry

Reply via email to