perl 5.18.0 status update

Front page | perl.perl5.porters | Postings from March 2013

perl 5.18.0 status update

Thread Next

From:

Ricardo Signes

Date:

March 12, 2013 01:47

Subject:

perl 5.18.0 status update

Message ID:

20130312014655.GC12799@cancer.codesimply.com


We're chugging along toward Perl 5.18.0.  There are still ten blockers.

Also, guess what?  It's time for us to start taking this code freeze seriously.
Please slow down on changing blead unless it is a clear, simple bug fix.  That
doesn't mean stop working on cool stuff... just keep it in another branch for
now.  The next task is to get blead smoothed out and released.

Perl 5.17.10 is due on March 21st.  (That's a day later than the usual 20th of
the month.  It is not a sigifnicant or interesting change, just a matter of
convenience for Max.)  After that, I'd like to see the rate of change slow
down.

Here are blockers, which I list in something like a rough priority order:

### 1.  Hash Randomization
    JSON::PP - https://rt.perl.org/rt3/Ticket/Display.html?id=116825
    EU::MM   - https://rt.perl.org/rt3/Ticket/Display.html?id=116857
    the rest - https://rt.perl.org/rt3/Ticket/Display.html?id=115908

Yves has a final bout of changes to hash randomization to land.  I haven't
spoken with him in a few days, but my understanding is that the only thing
blocking this landing is (a) some rebasing and squashing and (b) applying some
patches to two upstream-CPAN dists that haven't made new releases.

As far am I am concerned, this should land as soon as it's ready, so we can get
it even more tested.

### 2.  split "\x20" does not work as documented
    problem - https://rt.perl.org/rt3/Ticket/Display.html?id=116086
    cure    - https://rt.perl.org/rt3/Ticket/Display.html?id=116911

I would love to see this fix in, because I think it makes things a lot more
sensible.  Unfortunately, the branch with the cure seems to have hung Tony's
smokers a few times.  Needs eyes on.

### 3.  can't use @var in regexp, only $var
    https://rt.perl.org/rt3/Ticket/Display.html?id=115004

This is one of those "changes weird stuff that you'd think nobody is using, but
ha!  they totally are!" bugs.  A few weeks ago, Dave M. said this was on his
"things to look at before 5.18."  Great!  This needs to be addressed.

### 4.  Regexp::Grammars broken since 5.17.1
    https://rt.perl.org/rt3/Ticket/Display.html?id=116823

Same category as #3.  It sounds like "what Regexp::Grammars is doing shouldn't
work, but what they should be doing instead doesn't work either."

I am not worried about breaking Regexp::Grammars in 5.18.0, if Regexp::Grammars
can be fixed to work under 5.18.0.  If it cannot, and has to wait for 5.20.0,
then I am worried.

### 5.  add warnings for things that may be endangered
    ~~     - https://rt.perl.org/rt3/Ticket/Display.html?id=116913
    EBCDIC - https://rt.perl.org/rt3/Ticket/Display.html?id=116915
    WinCE  - https://rt.perl.org/rt3/Ticket/Display.html?id=115118

The EBCDIC and WinCE warnings can, if needed, live entirely in perldelta.

~~/smartmatch warnings cannot, so easily.  If we are going to radically change
these features, or remove them, we should really be warning.  We could just put
warnings in delta, but it would be nice to give users a more aggressive
warning of the instability of these features which have been in there for
several years now.

### 6.  "goto &sub" changes break Sub::Call::Tail
    https://rt.perl.org/rt3/Ticket/Display.html?id=115772

I'm willing to guess that Sub::Call::Tail is doing some nasty stuff here.
Still, it would be good at least to establish whether the module needs updating
(or maybe only its tests) or whether blead has been rendered broken in some way
that will bite us harder, later.

### 7.  Data::Util broken by changes to @ISA/overload interactions
    https://rt.perl.org/rt3/Ticket/Display.html?id=116088

Same as #6.

### 8.  documentation for THINKFIRST needed
    https://rt.perl.org/rt3/Ticket/Display.html?id=116925

We rolled back new-COW-by-default in part because we had not yet documented
THINKFIRST, which XS authors should sometimes be using.  There is also some
question of type maps.

We should get the documentation out in 5.18.0 so that there's more visibility
of the docs on how to do things safely, since we still want new-COW-by-default
in 5.20.0, right?

I don't know what THINKFIRST and the typemaps are all about.  This is on
somebody else.

### 9.  "in cleanup" warnings not enabled by "use warnings"
    https://rt.perl.org/rt3/Ticket/Display.html?id=113794

The bug's what it says on the tin.

### 10. __SUB__ should be undef in regex code block
    https://rt.perl.org/rt3/Ticket/Display.html?id=113928

This one, too.  Actually, it seems like "we're not sure exactly what should be
happening exactly in every way, but for now undef will do the trick."

### 11. Did I say 10?  Well, I meant 11.  perldelta!

It's time for me (starting tomorrow or so) to really start work on
perl5180delta.  Woo!  That, I hope, will be my main perl.git occupation between
now and release.

-- 
rjbs

Thread Next