[sane-devel] Proposal: SANE standard addition

Henning Meier-Geinitz henning@meier-geinitz.de
Thu, 26 Sep 2002 23:45:37 +0200


On Thu, Sep 26, 2002 at 11:09:20PM +0200, abel deuring wrote:
> >I think we can do this in SANE 1. It's not really a change but an
> >addition of a missing definition. We didn't even increase the minor
> >version for such changes in the past. Maybe we should?
> well, I don't mind much about the standard version numbering ;) But you 
> wrote in an earlier mail that some frontends may already support the 1 
> bit RGB mode.

xscanimage once did but currently segfaults :-( I think quiteinsane
also does. Once we change this I will update the test backend and
xscanimage and will tell the quiteinsane author.

> Since we should at least change the output of the test 
> backend to conform with the addition to/clarification of the standard, 
> the frontends should have some way to detect this change. So we should 
> perhaps add something like
>   "Before Sane release 1.0.9 RGB 1 bit data was stored in the reverse
>   order."

I don't think that this is a good idea because the standard describes
the theory and in theory it was undefined. The SANE standard doesn't
even know a "SANE release". Just because some implementation
chose to use it one way doesn't make it defined. So if anyone is
really interested, he can read older standards. That's why I will
increase the version number :-)

If it's really necessary to point out that it was undefined, we should
point to the old version number of the standard.