Re: [PATCH] flex: updated to 2.5.37

[Phil Blundell <philb@brightsign.biz>]

On Fri, 2013-05-10 at 11:46 +0100, Phil Blundell wrote:
> On Fri, 2013-05-10 at 11:32 +0100, Richard Purdie wrote:
> > On Fri, 2013-05-10 at 12:20 +0200, Martin Jansa wrote:
> > > On Fri, May 10, 2013 at 01:15:10PM +0300, Bogdan Marinescu wrote:
> > > > Removed the patches to flex, because they are now applied to the
> > > > latest version.
> > > 
> > > Can you check if 2.5.37 still reports flex.real when you call
> > > flex --version?
> > > 
> > > If yes, please integrate
> > > http://lists.linuxtogo.org/pipermail/openembedded-core/2013-May/039425.html
> > > I'll send v2 with added ${S} in a minute
> > 
> > I'm really not convinced we should be hacking flex to fix this. Can't
> > whatever parsing is being done to extract the version be a little more
> > relaxed about the name?
> 
> Or we could patch create_wrapper() to put the original binary in a
> different directory (but keep the same leafname) rather than renaming it
> to foo.real.  That would also address Martin's other concern with flex++
> not knowing its own name.

Or you could replace the shell wrapper that create_wrapper() currently
generates with either a compiled binary or a Python script.  That would
allow you to execl("/path/foo.real", "foo", ...), which I don't think
you can do in shell but which would avoid this whole class of problem.

p.