On Thu, 28 Feb 2008 08:43:26 +0900 Olaf Meeuwissen <olaf.meeuwissen at avasys.jp> wrote:
> Alessandro Zummo <azummo-lists at towertech.it> writes: > > > please see the attached patch regarding my proposal > > for a safe enabling of 1.1 features in the backends. > > Hmm, the SANE Standard Version 1.03 says about SANE_VERSION_CODE(): > > Note that the major version number alone determines whether a > frontend/backend pair is compatible. The minor version and the > build revision are used for informational and bug-fixing purposes > only. > > This means that a backend implementer is free to pass back something > like > > SANE_VERSION_CODE(1,2,3) > > for a SANE 1.0 backend. Isn't that a bit confusing? I think that > just changing the name of the macro to something like SANE_API_LEVEL > would improve code readability and reduce confusion. yes, SANE_API_LEVEL sounds better. The 1.1 standar will have to mention this. > > The same SANE Standard Version 1.03 does not explicitly state what a > backend is supposed to return when passed and unknown action. I'd > assume that any sane developer would return SANE_STATUS_UNSUPPORTED > but I wouldn't quite bet on it. neither do I, that's why I will be checking every backend. UNSUPPORTED would be fine but any error would do. The 1.1 standard will mention that too. > Another issue I have with the SANE_ACTION_COMPAT_LEVEL symbol is that > I don't think of that as an action. A symbol more action-like, like > SANE_ACTION_CHECK_API_LEVEL for example, is clearer IMHO. ack. > Apart from the above (and the fact that it is an ugly hack as SANE 1.0 > surely didn't intend to provide for API major/minor versions ;-), the > above would work fine in practice. We all noticed that SANE 1.0 missed some things.. that's the reason we are still discussing all of that 10 years after :-D -- Best regards, Alessandro Zummo, Tower Technologies - Torino, Italy http://www.towertech.it
