Re: 2.17.26 regression tests

[David Kastrup]

"Keith OHara" <address@hidden> writes:

> The test 'parenthesize.ly' now fails to meet its self-description, but
> that was due a change to the test input by David, so it seems to be
> one of his "not-yet-regression" tests that will pass after the patch
> for issue 1523 goes in.

Yes, sorry.  I thought it would be good to see the effect of the change
by having the regtest cover this as well.  Did not expext issue 1523 to
take that long to get in (almost a week, yes I know I am impatient).

> We have a release with no known regressions.

You make it sound as if issue 2910 will warrant a bunch of fine-tuning
before the results can be called production-ready.  That means that we
have an unfinished new state, and if I'm not mistaken, a state before
the change that is rather close to 2.16.  If we want a reasonably
straight progression of output results from 2.12-2.14-2.16-2.18, it
would appear that if we are aiming for a timely release, we have pretty
much reached the point where it makes sense to fork the stable branch
and start reverting stuff that is not going to make it, and
cherry-picking upcoming fixes from the unstable branch that are desired.

It's really a nuisance that backend changes tend to have long-ranging
consequences.  Partly (like with 2910) because they just affect a lot,
partly (like with the oodles of circular dependencies stuff) because the
design of the backend makes it just too fragile for making
straightforward changes with straightforward consequences.  As it
stands, neither the changes nor the consequences are straightforward.

-- 
David Kastrup