Re: [boost] Major code changes deadline approaching

Subject: Re: [boost] Major code changes deadline approaching
From: Stefan Strasser (strasser_at_[hidden])
Date: 2009-12-27 14:35:39

Am Sunday 27 December 2009 22:16:00 schrieb Vladimir Prus:
> > I'd just like to point out that we have
> > 28 open tickets marked as
> > "Showstoppers":	https://svn.boost.org/trac/boost/report/29
>
> Well, given that half of them is more than 6 months old, they cannot be all
> true show stoppers. Unfortunately, the severity of the issue typically
> represents submitter opinion, and not maintainer position about priority of
> a fix. That said, of course it would be nice if everybody looked at that
> list and try to do something.

I looked into the regressions.
I did not try to confirm or fix the bugs, but only put together the 
information that is in the reports and the changelogs from the trunk.

almost half of them can be closed or the regression flag removed imho, but I 
did NOT make any changes to Trac, as I don't want to tinker with release 
critical bugs before a deadline, although I'm pretty sure about some of them.

here we go:

Confirmed regressions:
(based on information in reports only, not tested)

#2409: 1.35 -> 1.36 optional
#2649: 1.34 -> 1.36 range
#2650: 1.36 -> 1.37 test (documentation only)
#3318: 1.36 -> 1.39 python
#3526: 1.34 -> 1.37 thread
#3668: 1.38 -> 1.41 intrusive
#3688: 1.38 -> 1.41 intrusive

Fixed, but still open in Trac:

#3010: iostreams. fixed.
fixed according to author.

#2154: iostreams. probably fixed.
the function in question(futimes()) is not in the trunk any more.
log from Rev 57607 that removed it: "This file is dead and is the only 
remaining place mentioning the futimes function mentioned in #2817."
but 2817 is an unrelated report that does not mention futimes(). report 
numbers mixed up?
author is danieljames.

Regression flag questionable:

#3178: thread.
no mention of a regression. can't be a (recent) regression: the reporter is 
talking about the win32 implementation("interruptible_wait"). the report is 
from 2009, the last changes to the relevant code in the trunk are from 2007.

#3670: asio.
was reported 5 weeks ago, with the following comment:
"I am pretty sure that this behaviour has been present since at least boost 
1.37.". no mention of a regression. mistake by the reporter?

#3392: test (documentation only)
report doesn't mention a regression, only that the bug is there "at least 
since 1.36". 1.36 is the oldest documentation that is online. mistake by 
reporter?

Not a bug:

#2107: MPL. someone included an external library that #defines the token "P4", 
which breaks MPL. boost libraries can't possibly work around the pollution of 
every other library out there.

Under investigation:

#3123: Serialization. author looking into it.