On Monday, 18 August 2014 at 21:57:19 UTC, bearophile wrote:
Vladimir Panteleev:
I agree, I am also surprised that 2.066 was released despite
the regressions.
There is an apparently endless stream of regressions, I have
found another today
(https://issues.dlang.org/show_bug.cgi?id=13321 ). I think D is
not yet at the stage of its development where it can hope to
fix all the regressions. So if you try to wait for all
regressions to be fixed, you never ship a compiler version, and
this has serious disadvantages. So better to be a little more
practical for now. 2.066 has took ages to come out, it was
overdue. I hope 2.067 will come out much quicker.
Bye,
bearophile
I have checked the regression list daily since something like b3
- amount of "hard" regressions was steadily going down and many
of newly added one were trivial and fixed quickly. Last time I
checked there were only 2-3 really problematic cases (including
one I have mentioned).
Idea is quite simple - if we are incapable of doing compiler
release without regressions, we should stop doing compiler
releases until we learn how to do it. Risk of reputation damage
we may get with 2.066 costs much more than delaying release even
for several months. Remember, we are speaking about regressions,
not even about critical bugs.
I also propose to start 2.067 beta branch right now and declare
it yet another bug-fixing release.