On Tue, 4 Feb 2003, Jeff Garzik wrote:

>> Changing the "(???)" to something else is also an option, but I'd 
>> prefer to do one of the above instead, and just wondered what 
>> common consensus might be.
>
>hehe... ran into this in the Linux kernel, too, where comments
>containing "???" occur from time to time :)
>
>ANSI trigraphs are pretty damn useless these days.  However, with
>the large number of compilers XFree86 must support, I think the safe
>bet is to be fully standard compliant -- including these annoying
>trigraphs nobody wants.  Why?  The alternative is to research how to
>disable trigraphs on _every_ compiler that XFree86 may support  :/

That's a very good point Jeff, one that hadn't dawned on me.

>I am only one voice without much weight,
>but I think we are cursed to live with trigraphs enabled...

Indeed, that does make the most sense.  It is probably as rare 
that someone would encounter this problem as it would be for 
someone to use trigraphs in C source code nowadays.  ;o)

>So, patching comments to not use trigraph sequences may be the only solution.
>Escaping is just kinda pointless when you could just change the string
>to be acceptable to humans _and_ ANSI-compliant compilers.
>
>There is always s/???/WTF?/ for example ;-) ;-)

Well, in this case, the driver displays "(AGP)" or "(PCI)" for 
different video hardware.  There are chips that I added support 
for which I am not sure wether they are AGP or PCI, so I put 
(???) until they can be rectified with documentation or user 
feedback.  So, no real harm done.  ;o)

The driver additions are harmless as far as working hardware is 
concerned, but my intention was to get other hardware working 
that isn't supported in the sources currently, but which should 
work without any driver changes - just conditional codepaths.  
I suppose I could just leave the (xxx) out entirely also, but 
then I couldn't keep tabs on which ones are known and which ones 
are unknown as easily either.  ;o)

I've changed it to "(\?\?\?)" now to workaround the trigraphs 
enabled issue, which is probably the sanest solution for now.

Thanks.


-- 
Mike A. Harris


_______________________________________________
Devel mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/devel

Reply via email to