Dear release team, I've messed up royally with the current upload of the package gavodachs2-server -- I've tested it in a buster container rather than an unstable container, and thus it was tested against python 3.7 and pyparsing 2.2.
This leads to bugs #984555 (due to python 3.8; this would be fixable with a patch of ~ 10 lines) and #985396 (due to pyparsing 2.4; this is the bad one, requiring embedding pyparsing 2.2 into gavodachs2). Fixing these requires something with a humonguos debdiff, which I've put up at https://docs.g-vo.org/dachs2.debdiff for now, as I'm really not sure if it's appropriate to attach this kind of thing to an unfreeze bug. Is there any chance to get that into bullseye? You see, the package as it's in now really is basically useless for what people will want to do with it. I've thought hard about the alternative, forward-porting to pyparsing 2.4, but that is a *major* undertaking; the grammars that are affected have hundreds of symbols, and I estimate about 200 non-trivial changes would be necessary to make things work. Getting this right on the first shot is certainly going to be hard; the pyparsing 2.2-based code, on the other hand, has matured over ~10 years (and a few previous pyparsing versions), so I'd *really* like to go carefully on this. Now, you can rightfully say "why didn't you properly configure the CI on salsa"? Mea culpa. But since you need to run a database server to enable meaningful tests, I've always relied on doing this in local containers. Perhaps if I do penance and work things out for salsa CI (hint welcome) -- could that move your hearts to letting in the patch? Sorry for the extra work, Markus
