Re: NSS support quirks
Date: Mon, 16 Jun 2008 18:19:13 +0200 (CEST)
On Mon, 16 Jun 2008, Axel Tillequin wrote:
>> Both these should be using <> and not "" since the headers won't be in the same dir as the .c file, right?
> well using <xxx> or "xxx" is not the real problem here. Strictly speaking I
> would say that <> are for standardized headers (ISO, POSIX) found in
> standard places (/usr/include in unix) while "" are used for other headers.
> This makes things clear but is probably too strict... (reminds me some
> emacs/vi discussions with friends...)
No. <> and "" have actual different functions as they work differently and
they should thus be used to make the code compile as intended. IMO. Using them
to differentiate "private" headers from "standard" is doomed to fail.
(Although I know a large portion of the world is still trying to use them that
> yes: with <base64.h>, ../lib/base64.h is found before
> /usr/include/nss/base64.h...too bad right ?
Yeah, supporting to build outside of the source tree gives that of course...
But isn't the nss headers always kept in its own subdir? They certainly are in
my installation. So can't/shouldn't we access _all_ the NSS headers with
nss/[header] to avoid these kinds of problems?
> The good thing with having nss/base64.h is that it makes it clear from the
> source what interface you are talking about.
-- / daniel.haxx.seReceived on 2008-06-16