From @simoncozens

Tom Christiansen (lists.p5p)​:

It is? Really? You're calling a working, dramatically improved,
and far more than simply timely release an EMBARRASSMENT?

Okay. I'm now sick of the bullshit.

First, let me say I don't want to take anything away from Sarathy. He's
done an *excellent* job as pumpking, and I can't praise him enough. This
release has needed a lot of nurturing, and he has given time and energy
above and beyond the call of duty to do exactly what has been required
at all times. Sarathy, thank you.

But one of the things I've wanted to do is tell the world what's new in
Perl 5.6.0. I've had quite a good crack at it, but believe me, it's been
extremely hard.

What improvements can we tell the user community about? To justify a
move off the 5.5 track, it's got to be something big.

Ah, yes. Unicode. But after two years of work, the one thing that users
will want to do - open and read Unicode data - is still not there.
Who cares if stuff's now represented internally in Unicode if they can't
read the files they need to.

The compiler dumps core less often. I could say that. I don't think
it'll give the right impression.

Threading. We've got this brand new threading model but - sorry folks,
you can't actually use it. The only thing it does at the moment is give
fork() to those systems which don't have it. The only significant group
of users this affects are Windows users​: ActiveState's customers.

This shouldn't be an issue, as there are other PC ports around. But support
for Cygnus now requires a development snapshot to work. DJGPP people
have been reporting failures as well. This may be due to inadequacies in
DJGPP - which we failed to work around. The only PC port that's held it
together has been - guess.

Hey, don't draw the wrong conclusion here - this just means Sarathy has
been working harder and better than the rest of us. But it's still
pretty odd. Especially since we've allowed things like AIX to lapse as
well.

Those are the three big things we've been working on. They're not there.

No, don't parrot me the change log. I've read it. I've crawled through
perldata and Changes to fish out things worth writing about. It's not
been easy.

I've also read Todo-5.6, and seen everything we said we would do, and
somehow the release became `ready' before we made good on a reasonable
number of those. That worried me.

But besides, for the purposes of selling a new version to the users,
I've been trying to find things that'll be significant for them, not us.

Take Unicode tuples. Great stuff, but they've caused a whole bunch of
confusion among the porters - the very people who ought to understand
this stuff better than anyone else. If we can't grok it, I can't really
push it as new and exciting.

Yeah, a bunch of bugs have been fixed. So, what, I'm reduced to calling
it a bugfix release? But that would have been 5.005_04.

Eventually, I cooked something plausible up, but had to drown it in
provisos and disclaimers enough to take all the wind out of it. It's
been depressing.

We were overdue a new release. I accept that. And it's merely a new
subversion, not a brand new version. Fair enough.

I also accept this wasn't Sarathy's fault. If it was anyone's fault, it
was my fault, because I didn't work hard enough to get the things I
cared about on the Todo list done. I hope each one of you can honestly
say the same.

But this means I won't be getting the champagne out yet.

And I means I seriously don't think it was worth three bloody good
porters over.

From [Unknown Contact. See original ticket]

If this hadn't gotten out NOW, it might well NEVER would have gotten
out? Why? Because Sarathy would have quit in frustration over the
demands put upon him, the expectation that he and he alone would
rewrite and redesign all the major subsystems for everybody's pet
little feeps, from threads to unicode to the codegenerators. That's
asking a hell of a lot of one *volunteer*.

We don't *need* those features for there to be a new release. Plenty
got done. I challenge you to do back and look at what really
happened for 5.N for other values of N in the (1,2,3,4,5) range.
5.002 says "prototypes". 5.003 is marked as a "security release".
5.004 is marked as a "maintenance release".

You don't want perl5.6 -- you want perl6. That's a fine thing to
want. But by any historical metric, this release had plenty in it
to count as a dot-N release, and two years is too long to wait for
this. Release early, release often.

I see you're all lining up to take the switch to Larry's backside.
How thoughtful.

--tom

From [Unknown Contact. See original ticket]

On 1 Apr 2000, Simon Cozens wrote​:

Especially since we've allowed things like AIX to lapse as
well.

Lapse in what way? I labored long and hard to get LP64 model builds
working reliably on newer RS/6000 hardware, and it's used in production
daily at my employer's shop.

The only open issue that I'm able to confirm relates to socket misbehavior
on AIX 4.2.x (32-bit platforms). Neither Jarkko nor I saw any problems
prior to release, so I'm not sure how this slipped through. It's my
intention to nail the problem in the near future (though fixes from the
community would be fine).

Steve

From [Unknown Contact. See original ticket]

Simon Cozens writes​:

This shouldn't be an issue, as there are other PC ports around. But support
for Cygnus now requires a development snapshot to work. DJGPP people
have been reporting failures as well. This may be due to inadequacies in
DJGPP - which we failed to work around. The only PC port that's held it
together has been - guess.

Well, lemme try... OS/2? ;-)

Ilya

From @gsar

On 01 Apr 2000 10​:32​:22 GMT, Simon Cozens wrote​:

Okay. I'm now sick of the bullshit.

Me too, but I think you could have guessed that without my saying so. :-)

First, let me say I don't want to take anything away from Sarathy. He's
done an *excellent* job as pumpking, and I can't praise him enough. This
release has needed a lot of nurturing, and he has given time and energy
above and beyond the call of duty to do exactly what has been required
at all times. Sarathy, thank you.

You're welcome, I'm sure.

What improvements can we tell the user community about? To justify a
move off the 5.5 track, it's got to be something big.

"Big" is in the eyes of the beholder. 5.6.0 is "big" in various ways.
If you choose to ignore them, I guess you get what you choose.

Besides, I think making decisions based on whether something is "big"
or not is a lousy way to make decisions.

Ah, yes. Unicode. But after two years of work, the one thing that users
will want to do - open and read Unicode data - is still not there.
Who cares if stuff's now represented internally in Unicode if they can't
read the files they need to.

This is a "big" (as in "huge") disappointment for me as well. I hope
we'll do better next time.

The compiler dumps core less often. I could say that. I don't think
it'll give the right impression.

Why not? I think we _do_ want to give the correct impression, whether
we feel good about it or not. The fact is that the compiler isn't quite
there yet, and we had better admit it. (I might say I consider it a
failure of the community that it isn't "there" yet, but that isn't a very
productive observation.)

Threading. We've got this brand new threading model but - sorry folks,
you can't actually use it. The only thing it does at the moment is give
fork() to those systems which don't have it. The only significant group
of users this affects are Windows users​: ActiveState's customers.

Er, not all Windows users are ActiveState's "customers". In fact, a tiny
proportion of the people that use ActivePerl are ActiveState's customers,
in the sense that they don't pay a thin red cent for what they get (and
often, take for granted).

And I should add that Windows users _are_ a significant portion of the
Perl user community. (You seem to think otherwise in that last sentence.)

This shouldn't be an issue, as there are other PC ports around. But support
for Cygnus now requires a development snapshot to work.

This was a choice made by the Cygwin port maintainers, and I don't quarrel
with them if they think that's the best choice.

                                           DJGPP people

have been reporting failures as well.

I don't think this is entirely true. Laszlo Molnar, the maintainer
of the DJGPP port, reported success. That's why the announcement had an
entry for DJGPP.

                                  This may be due to inadequacies in

DJGPP - which we failed to work around. The only PC port that's held it
together has been - guess.

If you're insinuating that the Windows port is the only one that ever gets
maintained, you'd be judging all the people who worked on the other
ports rather harshly.

Hey, don't draw the wrong conclusion here - this just means Sarathy has
been working harder and better than the rest of us. But it's still
pretty odd. Especially since we've allowed things like AIX to lapse as
well.

This also appears to be incorrect information. I will be charitable and
not call it FUD. :-)

But besides, for the purposes of selling a new version to the users,
I've been trying to find things that'll be significant for them, not us.

I don't "buy" the notion that every new version of Perl has to be "sold"
to users. Tell them like it is, and let them judge. (I trust perldelta
and the announcement did that adequate justice.)

Take Unicode tuples. Great stuff, but they've caused a whole bunch of
confusion among the porters - the very people who ought to understand
this stuff better than anyone else. If we can't grok it, I can't really
push it as new and exciting.

Maybe you shouldn't "push it" at all then. Maybe you're just looking
for what's wrong rather than what's right, and that might also make you
the wrong person to be "pushing it".

We were overdue a new release. I accept that. And it's merely a new
subversion, not a brand new version. Fair enough.

I'd say this is mere semantics (as in "it's all in your head") but I
guess you could retort with "so is everything else under the sun!"

I also accept this wasn't Sarathy's fault. If it was anyone's fault, it
was my fault, because I didn't work hard enough to get the things I
cared about on the Todo list done. I hope each one of you can honestly
say the same.

I'm perfectly willing to shoulder any portion of the blame for 5.6.0, as
long as there is fairness in it. So far, I haven't seen any evidence
of that.

Maybe being unfair is in fashion these days.

Sarathy
gsar@​ActiveState.com

From @samtregar

On 1 Apr 2000, Simon Cozens wrote​:

Yeah, a bunch of bugs have been fixed. So, what, I'm reduced to calling
it a bugfix release? But that would have been 5.005_04.

Whoa. It's not "Perl 6". The numbering convention changed, and if you
can buy that then it's only a .1 change. It's got a LONG changelog for a
.1 change, at least in my book.

What exactly is the point of all this bitching? Do you want to rewrite
the past? Are you volunteering to make sure 5.7 is totally awesome?

-sam