[sane-devel] [sane-commit] CVS update of sane-backends (37 files)

[Chris Bagwell]

On Wed, Feb 25, 2009 at 11:08 PM, stef <stef.dev at free.fr> wrote:

>
>
>         Hello,
>
>        I would have preferred to ear of that change before. It comes at a
> wrong time
> since I'm doing and testing a non-trivial merge in the genesys backend.
>
>        I didn't see any warning of such a global change. Did I miss an
> email ?


Sorry about that... I bounced the idea off a few people but didn't submit to
mailing list before I did it.  I think running the following commands should
allow you to recover relatively easy (also its the easy way to revert my
commit if it comes to it).

Run them should cause most the cvs update to say changes are already
applied.  Or at least it can be used to update your genesys files and let
you drop into a new copy of CVS to skip update step.

sed -i 's/u_int8_t/uint8_t/g' *.[ch]
sed -i 's/u_int16_t/uint16_t/g' *.[ch]
sed -i 's/u_int32_t/uint32_t/g' *.[ch]



>
>        However the 'right' type to use is an interesting point, I've never
> been
> really convinced with the mix of SANE_Int, int and u_int16 variables.


I'll not speak for SANE_Int as it may have a special purpose.  For the rest,
I usually don't mind to support a range of types that people are comfortable
with.  Its just the u_int*_t are a BSD-ism and have portability issues;
whereas uint*_t are now defined by C standard and should be easier to port
(although stdint.h does have a suprisingly large amount of baggage itself).

I purposely left in the u_char_t, u_int_t, and u_long_t since they are not
defined by same standard and do not have same portability issues for what
ever reasons.

Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.alioth.debian.org/pipermail/sane-devel/attachments/20090226/690f5161/attachment.htm