Hi, Nelson and Luk,
While browsing the 'net, I run across the thread
"Unblock qsf 1.2.6-1 and optipng 0.5.5-1" at
http://lists.debian.org/debian-release/2007/02/msg00711.html
(Nelson knows ;-) I am interested in the outcome of this discussion.
As the upstream author of OptiPNG, I would like to point out that 0.5.5
is a recommended release, because of some high-importance fixes that it
carries.
I understand that Etch is to be released soon, and, due to stability
concerns, only the security and critical bug-fix updates are allowed on
Testing. Since I am not so familiar with the Debian release policies,
I am unsure whether 0.5.5-1 will make it to Testing in due time for the
Etch release.
I don't want to rush things beyond what's considered "appropriate
incubation" for a Debian package, but I do wish to emphasize that
OptiPNG-0.5.5 carries, among other things, two important fixes.
Thanks to Nelson, one of them was already put in the Debian package
OptiPNG-0.5.4+0-1, but the other is only in 0.5.5, and it is important
because it fixes a potential loss of metadata (background color).
When I made the 0.5.5 release, I was careful to put in bug fixes only.
(There are more than just these 2 fixes, but the rest are minor.) Yet,
by reading the above-mentioned thread, I realize the volume of change
from 0.5.4 to 0.5.5 is still too large, and I wonder if there is a way
that would favor a more rapid transition to Testing. For instance, it
could be possible to have a new 0.5.4 patch that incorporates solely
the critical fix from 0.5.5. That can be achieved by incorporating
optipng-0.5.5/src/opngreduc.c directly into the new 0.5.4 (and create,
say, a package release named optipng-0.5.4+1-1). Such a file movement
is safe, because the changes from opngreduc.c do not have side effects
on the rest of the program.
Moving src/opngreduc.c from 0.5.5 into a hypothetical 0.5.4+1-1 would
lead to a release that consists of 1 file changed, 16 additions and 5
deletions. (There are 2 more additions and 2 deletions, but those ones
only change the amount of whitespace.)
I understand that making a new back-ported 0.5.4 release will add
another time overhead, because it needs to incubate inside Unstable.
On the other hand, if such a course of action would still lead to a
faster acceptance into Testing, and has a better chance to make it
into Etch than the current 0.5.5-1, then I recommend taking it.
Thank you very much for your attention.
Best regards,
Cosmin Truta