nanogui: Pending patches/bug/requests ... list available on the FTP


Previous by date: 1 Jun 2005 14:48:10 +0100 Alpha blending using Nano-X APIs, Alan Chuang
Next by date: 1 Jun 2005 14:48:10 +0100 Re: #if FOO vs. #ifdef BAR, Alain Volmat
Previous in thread: 1 Jun 2005 14:48:10 +0100 Re: Pending patches/bug/requests ... list available on the FTP, Alain Volmat
Next in thread: 1 Jun 2005 14:48:10 +0100 Re: Pending patches/bug/requests ... list available on the FTP, Jachym Holecek

Subject: Re: [nanogui] Pending patches/bug/requests ... list available on the FTP
From: Alain Volmat ####@####.####
Date: 1 Jun 2005 14:48:10 +0100
Message-Id: <20050601134803.GA10327@pop.mail.yahoo.com>

Hi,

since nobody have send any comment about this method (and I'm now updating
the patch_092 folder to add recent discussions/patch) I give my opinion
about it.

>   * At compile time, scan all the driver sources for DEFDRIVER() and have
>     a tiny sed+awk/whatever script construct a .c file with the driver
>     table.

I'm a bit worrying about such solution actually. Of course that's only
my feeling and others should also give their opinion on it.
The idea of the dynamically generated .c file is a bit frightening to me,
wouldn't it becomes a bit harder to really see what is being compiled ?

Regards,

Alain


* Jachym Holecek [Sat, 28 May 2005 at 18:15 +0200]
<quote>
> Hello,
> 
> > I'd like to see this one go in, I was about to start writing the same kind of 
> > patch. Another option would be to have all drivers register themselves.
> > So in each driver file something like
> > 
> > static void register_me() __attribute__ ((constructor))
> > {
> >    global_driver_table_append(&my_driver_struct);
> > }
> > 
> > Then the driver selection code in fb.c would turn from a big switch to an 
> > iteration over the list of drivers and choose the matching one. This would 
> > get rid of the ifdefs in this file.
> > 
> > Would you consider this a better option ?
> 
> In case portability is of concern, you can avoid the __attribute__ quite
> easily:
> 
>   * Have each driver contain "DEFDRIVER(my_driver_struct)", where DEFDRIVER
>     is a no-op macro.
> 
>   * At compile time, scan all the driver sources for DEFDRIVER() and have
>     a tiny sed+awk/whatever script construct a .c file with the driver
>     table.
> 
> That's what I do in a private project of mine, and it works reasonable
> well. If people think this would be a good way, I'd volunteer to implement
> it.
> 
> 	Regards,
> 		-- Jachym Holecek
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ####@####.####
> For additional commands, e-mail: ####@####.####
> 
</quote>

	

	
		
_____________________________________________________________________________ 
Découvrez le nouveau Yahoo! Mail : 1 Go d'espace de stockage pour vos mails, photos et vidéos ! 
Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com

Previous by date: 1 Jun 2005 14:48:10 +0100 Alpha blending using Nano-X APIs, Alan Chuang
Next by date: 1 Jun 2005 14:48:10 +0100 Re: #if FOO vs. #ifdef BAR, Alain Volmat
Previous in thread: 1 Jun 2005 14:48:10 +0100 Re: Pending patches/bug/requests ... list available on the FTP, Alain Volmat
Next in thread: 1 Jun 2005 14:48:10 +0100 Re: Pending patches/bug/requests ... list available on the FTP, Jachym Holecek


Powered by ezmlm-browse 0.20.