DocBook v2

[Pavel Sanda]

On Mon, Aug 31, 2020 at 03:49:44PM -0400, Scott Kostyshak wrote:
> On Mon, Aug 31, 2020 at 09:26:30PM +0200, Pavel Sanda wrote:
> 
> > If you are kind of finished with this part I would also suggest to
> > start pushing those changes to master and let other peoplee help
> > with the small stuff like the warnings from different compilers...
> 
> You did not mention anything about the ctests, so the following is not
> directly a response to your message, but for the ctests I would prefer
> to get them figured out before merging. The ctests often find
> regressions that take time to fix.  By having all that done on the
> branch, people can still test master with less fear of regressions
> slipping in; and the ctests continue being valid on master and can catch
> regressions on master. If the ctests are broken because there is a
> regression on master, and there are (as expected) commits in the
> meantime, those commits could break something that would not be caught
> by the tests.

I understand the general reasoning but
a) I find it unfair to force Thibaout to figure out tests on windows, so
   his stuff gets merged. The more that I expect bunch of weird issues
   popping up while trying to do so. While having tests working in 
   windows would be nice, if its not low hanging fruit I would like
   his free time to be primarily spent on docbook instead of figuring
   out perl configuration delicacies.

b) docbook export is completely new and we are not really breaking old
   legacy. My suggestion is that if failling docbook tests disturb you,
   simply disable all of them and start enabling them and slowly one 
   by one. It seems to me that having other people to test the code on
   various architectures and even release some technical versions of 2.4
   so people can give feedback to the actual output overweigths the tests
   issues.

(I might be wrong on this if you suggest that his code breaks non
dobook tests...)

Pavel