On 2012.02.20 12:45, Rocky Bernstein wrote:
> There have a few larger unrelated changes that taken place and I would like
> to solicit opinions on whether we should have one release with all of the
> changes or two?
> The changes are
> * CD-Text completion (some incompatibility)
> * UDF improvement and header reworking for Microsoft OS's
> * Removal of cd-paranoia which is be in now in a separate GPL v2+ directory
> * Various bug fixes
Note that on top of that, I have recently added some Joliet fixes (as
well as some addressing of the TODOs with regards to falling back to non
Joliet if the string is the same but potentially larger - see  and ).
The main issue was that some ISO9660 discs may have multiple Secondary
Volume Descriptors (eg: El Torito + Joliet) and the existing code only
handled the very first one for Joliet.
I also had to amend the fixes for MSVC UDF compatibility, as I
discovered that in the Microsoft world, an union of zero-sized arrays is
not zero bytes as expected (and as is the case for GNU), but one byte.
This created a problem with file entries as the MSVC sizeof was off (but
this was only an issue for MSVC compiled code). This only has an impact
for UDF , but renders the header workarounds required for MSVC
slightly more in-your-face, as we now need to attach the MSVC required
unions to a non-empty member.
Now, with regards to your question, my preference would be a single
release, as it'd obviously mean less work than having to split commits.
Also a lot of the UDF "improvements" are actually bug fixes for LFS or
MinGW, which are supposed to be already at least partially supported
with LFS fixes also applying to ISO9660.
It's very possible that if we try to have a first release that fixes as
many issues as it can, there might actually not be much content left for
the subsequent release...