Re: 5.6.0 ready for prime-time?

[p5pRT]

Migrated from rt.perl.org#2934 (status was 'resolved')

Searchable as RT2934$

[p5pRT]

From pete@petes-place.com

I\, for one\, am advising my users and readers to avoid 5.6 at this time. Although it does compile on Win32 right now\, it does not do so and remain stable for long. It does fail tests\, and from what I'm seeing the most important modules to Win32 users are not ready for 5.6 internals and therefore do not compile. Since I don't make it my business to meddle in the internals of add-on modules\, 5.6 is not usable to Win32 users at this time who need more than core perl\, which itself is very questionable.

As for why it was released early I have no idea\, but wouldn't mind seeing this thread following to fruition. I am seeing companies and groups deny pressuring Sarathy for an early release. This is quickly becoming an _embarrassment_ to this group and the Perl community\, and there seems no reason to doubt that this will worsen in a very short time. I also suspect that the culprit will emerge shortly as well.

There was also no warning that I've seen that 5.6 was an early beta\, or early-beta-ish\, though that's certainly where it's stands. The Win32 community is convinced that 5.6 is out\, stable\, and available only from one source. The only company I see making a big deal out of this is ActiveState\, and in their defence (surprise surprise) I don't see any logical reason why they would cut their own throats by hurrying something that just wasn't ready for release. I've tested AS's release with the precompiled modules available from them\, and am seeing all of the troubles that I'm finding in my independent release.

IS Perl 5.6 considered stable?

I would also qualify "stable" with the stability of the more popular modules\, Tk\, LibWin32\, LibWWW\, LibNet\, GD\, DB_File\, and other biggies\, many of which do not compile at all with 5.6 without "relatively minor tweaking". Without these modules\, Perl is not Perl. I am at the dawn of the first official release of CodeMagic\, Perl's only world-class editor at this time soon to go cross-platform\, which depends on a stable Perl as a backbone​: and after officially declaring 5.6 ready\, this group is now wondering why it was released at all so quickly? I have user who are going to question this\, and I don't have an answer for them.

David Grove pete@​petes-place.com

[p5pRT]

From @samtregar

On Fri\, 31 Mar 2000\, David Grove wrote​: I\, for one\, am advising my users and readers to avoid 5.6 at this time.

That's fine. People advising "users" very rarely advise an instant upgrade to a new and unproven version of a complicated program. The only exception is when the old version is unworkable - which 5.005_03 is certainly not.

The cautious are well served by never upgrading *any* software to a verison less than a few months old.

and from what I'm seeing the most important modules to Win32 users are not ready for 5.6 internals and therefore do not compile.

No surprise there. Module authors are really a lazy bunch when it comes to making patches for development versions. I maintain a few packages and only one of them works with 5.6! I'll be releasing 5.6 versions soon enough and I'm sure I won't be alone.

IS Perl 5.6 considered stable?

Yes\, I believe I've heard Sarathy state at least a few times that it is.
Among other things\, 5.6 is a bug fix release. 5.005_03 is not exactly bug free\, I hope you realize!

I am at the dawn of the first official release of CodeMagic\, Perl's only world-class editor at this time

That's a joke\, right? Have you heard of Emacs? That's my "world-class" Perl editor of choice.

I have user who are going to question this\, and I don't have an answer for them.

Repeat after me - "I recommend holding off upgrading for at least a month or two. A number of important modules don't work with 5.6 yet\, but they will soon."

Is that really so hard? Do you think it could have been any other way?

-sam

[p5pRT]

From [Unknown Contact. See original ticket]

That's fine. People advising "users" very rarely advise an instant upgrade to a new and unproven version of a complicated program. The only exception is when the old version is unworkable - which 5.005_03 is certainly not.

The cautious are well served by never upgrading *any* software to a verison less than a few months old.

True. This is a lesson I thought I'd learned. I didn't let my teams upgrade to Delphi 5.0 when it was released because "we don't know what bugs there are in it yet... it needs some time to mature and build a community of users". I wish I'd remembered that when I upgraded from NT to 2K. If I'd known that any slight change in network settings would require a reinstall... grrr... some security feature.

I'm not expecting perfection in the latest\, greatest release of Perl\, but I do think that the subcommunities that we represent would be better served by releasing after a real _public_ beta period\, longer than say a month\, with realistic time given to bring some of the modules up to date... at the very least the most important ones on a given platform\, like LibWin32 and Tk.

and from what I'm seeing the most important modules to Win32 users are not ready for 5.6 internals and therefore do not compile.

No surprise there. Module authors are really a lazy bunch when it comes to making patches for development versions. I maintain a few packages and only one of them works with 5.6! I'll be releasing 5.6 versions soon enough and I'm sure I won't be alone.

I wouldn't say lazy\, really\, just overworked\, tired and with more joy hacking and coding than maintaining; especially maintaining something when the rug is pulled out from under you. I'm wondering whether the term "backwards compatibility" has escaped us.

IS Perl 5.6 considered stable?

Yes\, I believe I've heard Sarathy state at least a few times that it is. Among other things\, 5.6 is a bug fix release. 5.005_03 is not exactly bug free\, I hope you realize!

I am at the dawn of the first official release of CodeMagic\, Perl's only world-class editor at this time

That's a joke\, right? Have you heard of Emacs? That's my "world-class" Perl editor of choice.

Heh\, yeah I guess it's a religious issue. But realize that CodeMagic is actually a Perl editor first and foremost\, built by (a) perl programmer(s) for perl programmers. The support for other languages just evolved as a natural counterpart of some free scripts that came with a third-party Delphi control. It's certainly not just a generalized editor as far as perl itself is concerned. It wasn't just thrown together like the host of Win32 editors like an overmarketed notepad.exe.

For ages Win32 folks have been screaming for a Win32 IDE/Editor for Perl. AS hasn't said an audible word about "Visual Perl" since they failed to deliver in spring\, summer\, fall\, and winter 98. (I eventually lost track of the promises.) Several general-purpose editors have _support_ for Perl\, but none are built upon and around it. None use Perl as its scripting/api/macro language. My greatest hope for it is that it be a catalyst for more corporate entities to embrace Perl because of it\, because\, if for no other reason\, it has buttons. ;-))

It's my baby and my contribution. It represents thousands of hours of work and contribution\, despite Tom's rant contrawise. Forgive a bit of earnest partiality.

I have user who are going to question this\, and I don't have an answer for them.

Repeat after me - "I recommend holding off upgrading for at least a month or two. A number of important modules don't work with 5.6 yet\, but they will soon."

Is that really so hard? Do you think it could have been any other way?

Point taken. Normally this is a given. This is unfortunately not true on Win32\, however\, where people are being told that 5.6 is stable and fork() works and all of this is brought to you by the kindness and generosity of [insert your favorite corporate-monopoly-want-to-be here].

I understand that it will take a while to catch the modules up. I'm anxiously awaiting that moment. At that point in time I can go on with my work and not have to deal with a deceived subcommunity hungry for a stable and complete Win32 port that fork()s and comes with stable modules.

[p5pRT]

From [Unknown Contact. See original ticket]

David Grove \pete@​petes\-place\.com writes​:

I'm not expecting perfection in the latest\, greatest release of Perl\, but I do think that the subcommunities that we represent would be better served by releasing after a real _public_ beta period\, longer than say a month\, with realistic time given to bring some of the modules up to date... at the very least the most important ones on a given platform\, like LibWin32 and Tk.

Okay - you mentioned Tk twice now. What exactly is your gripe about Tk and perl5.6.0 ?

The only issue that I am aware of with Tk and perl5.6.0 is that some long deprecated coding practices (e.g. not declaring the widgets you use) lead to peculiar error messages due to the ->isa bug. Tk800.020 fixes that. (It is worth noting that the 'use base'/->isa bug is a side effect of a patch contributed to the mailing list and not anything fundamental to the newness of 5.6.0 - just lack of test suite coverage and patch author's lack of knowledge of the side effects.)

I wouldn't say lazy\, really\, just overworked\, tired and with more joy hacking and coding than maintaining; especially maintaining something when the rug is pulled out from under you.

No rug was pulled from Tk - Tk is excessively familiar with perl's guts and despite that extemely minimal changes were required.

I'm wondering whether the term "backwards compatibility" has escaped us.

It is my watchword.

[p5pRT]

From [Unknown Contact. See original ticket]

Okay - you mentioned Tk twice now. What exactly is your gripe about Tk and perl5.6.0 ?

No gripes\, Nick. Your work is excellent and I applaud you\, period. You're one of the most active and effective members in this forum and have done tremendous work for the _entire_ perl community\, Win32 included.

(Side note\, you said a while back that you were thinking of making a MinGWin32 (GCC-Win) makefile for perl. Did that come about? If not I'll grab that task if you're willing to give it - I'd prefer that CM for one make its own Perl and not depend on a $1500 compiler\, and that people\, given a short tutorial\, not have to depend on AS's handouts.)

The only issue that I am aware of with Tk and perl5.6.0 is that some long deprecated coding practices (e.g. not declaring the widgets you use) lead to peculiar error messages due to the ->isa bug. Tk800.020 fixes that.

Another update? I can't keep up. ;-))

Actually what I personally am talking about is the nv vs PL_nv set of bugs. I'm afraid I don't know enough of perlguts to know what's going on there. It seems to be the same problem present in the other modules that failed on me. Frankly\, I'm not aware of any other bugs in Tk->5.6.

I don't know the bugs that the others here are talking about. I haven't gotten that far in my testing. I wasn't expecting 5.6 to be released so soon after a first beta\, so it caught me off guard too.

I wouldn't say lazy\, really\, just overworked\, tired and with more joy hacking and coding than maintaining; especially maintaining something when the rug is pulled out from under you.

No rug was pulled from Tk - Tk is excessively familiar with perl's guts and despite that extemely minimal changes were required.

Given an adequate beta period\, would these changes/fixes have been made before 5.6's release?

I'm wondering whether the term "backwards compatibility" has escaped us.

It is my watchword.

I've noticed. Kudos.

[p5pRT]

From [Unknown Contact. See original ticket]

David Grove \pete@​petes\-place\.com writes​:

Okay - you mentioned Tk twice now. What exactly is your gripe about Tk and perl5.6.0 ?

No gripes\, Nick. Your work is excellent and I applaud you\, period. You're one of the most active and effective members in this forum and have done tremendous work for the _entire_ perl community\, Win32 included.

Blush!

(Side note\, you said a while back that you were thinking of making a MinGWin32 (GCC-Win) makefile for perl. Did that come about? If not I'll grab that task if you're willing to give it - I'd prefer that CM for one make its own Perl and not depend on a $1500 compiler\, and that people\, given a short tutorial\, not have to depend on AS's handouts.)

Mingw32 has been supported since 5.005 - though I am planning to take another look with latest Mingw32 (gcc-2.95.2).

Actually what I personally am talking about is the nv vs PL_nv set of bugs.

Tk has supposed to have been immune to those for a while. If you have a build which shows them I am interested in fixing them.

I'm afraid I don't know enough of perlguts to know what's going on there. It seems to be the same problem present in the other modules that failed on me. Frankly\, I'm not aware of any other bugs in Tk->5.6.

No rug was pulled from Tk - Tk is excessively familiar with perl's guts

and despite that extemely minimal changes were required.

Given an adequate beta period\, would these changes/fixes have been made before 5.6's release?

With the exception of the ->isa bug they were made before 5.6's release. The ->isa bug fix was made before 5.6.0 was announced but after it reached CPAN.

[p5pRT]

From [Unknown Contact. See original ticket]

I've found it quite stable and healthy. I'd tried replacing the 5.005_03 that came on Red Hat with 5.005_63\, and had to back out of that since sdf (the Simple Document Format processor) blew chunks all over the place\, couldn't format anything. 5.6.0 is working fine; no problems with sdf\, and I use lots of other modules in other scripts\, and they're all happy too. Repackaging all the modules I use (with my perlmod2rpm script) also worked fine.

-Bennett

[p5pRT]

From [Unknown Contact. See original ticket]

That's a joke\, right? Have you heard of Emacs? That's my "world-class" Perl editor of choice.

Heh\, yeah I guess it's a religious issue. But realize that CodeMagic is actually a Perl editor first and foremost\, built by (a) perl programmer(s) for perl programmers. The support for other languages just evolved as a natural counterpart of some free scripts that came with a third-party Delphi control. It's certainly not just a generalized editor as far as perl itself is concerned. It wasn't just thrown together like the host of Win32 editors like an overmarketed notepad.exe.

For ages Win32 folks have been screaming for a Win32 IDE/Editor for Perl. AS hasn't said an audible word about "Visual Perl" since they failed to deliver in spring\, summer\, fall\, and winter 98. (I eventually lost track of the promises.) Several general-purpose editors have _support_ for Perl\, but none are built upon and around it. None use Perl as its scripting/api/macro language.

Perlmacs 0.14 is now making its way through CPAN.

  $CPAN/authors/id/JTOBEY/Emacs-PerlmacsPatch-0.14.tar.gz   or   http​://john-edwin-tobey.org/perlmacs/src/perlmacs-0.14.tar.gz

It hasn't been ported to Windows\, though.

My greatest hope for it is that it be a catalyst for more corporate entities to embrace Perl because of it\, because\, if for no other reason\, it has buttons. ;-))

Ah! Forgot to include buttons...

It's my baby and my contribution. It represents thousands of hours of work and contribution\, despite Tom's rant contrawise. Forgive a bit of earnest partiality.

Ditto. :-)

[p5pRT]

From @samtregar

On Sat\, 1 Apr 2000\, David Grove wrote​:

Heh\, yeah I guess it's a religious issue. But realize that CodeMagic is actually a Perl editor first and foremost\, built by (a) perl programmer(s) for perl programmers.

Emacs cperl-mode. A perl mode built by perl programmers for perl programmers. And it works a hell of a lot better than any "IDE" I've ever used. I'll resist the urge to slander your editor sight unseen\, but lets just say I'm anxious to get a look at this paramount effort.

My greatest hope for it is that it be a catalyst for more corporate entities to embrace Perl

That is absolutely hideous. What about "corporate entities" makes you desire their embrace? I'd prefer they stayed at arms length. Pass the cash through bullet-proof glass\, that sort of thing.

At that point in time I can go on with my work and not have to deal with a deceived subcommunity hungry for a stable and complete Win32 port that fork()s and comes with stable modules.

Do you have a single shred of evidence of willfull deception? Do you have anything indicating that Sarathy even once declared that 5.6 would be released totally stable and that CPAN would be instantly compatible?

I find these baseless claims rude and disrespectful to some people that I hold in high esteem.

-sam

[p5pRT]

From [Unknown Contact. See original ticket]

My greatest hope for it is that it be a catalyst for more corporate entities to embrace Perl

That is absolutely hideous. What about "corporate entities" makes you desire their embrace? I'd prefer they stayed at arms length. Pass the cash through bullet-proof glass\, that sort of thing.

No no no.

I'm not talking about corporate interference or selling out like [...] is trying to do with perl. I just mean that\, since I began actively advocating Perl in 95\, one of the most common corporate responses is "Where's the buttons?" This is reflected in the ubiquitous newbie question "How do I run a program?" or "Where's the Editor?". This has been a stumbling block for Perl in the Win32 community.

By providing the IDE interface\, the buttons to push\, the place to type\, the help in a Win32-normal help system\, and a standard port of Win32 perl\, I'm hoping to make it more attactive to people in charge. If the people in charge see perl as a complete system rather than as an "old\, worn out unix method" (quoting my previous supervisor\, whose job I now hold)\, then exhausted programmers can go home to their families at 5pm rather than staying until midnight every night trying to do perl's job in visual basic or c and fighting with the microsoft support systems.

Old marketing surveys from 96-97 showed that companies were moving from Novell and Unix to NT because they thought they had no choice. Since that time\, that choice has been a painful and expensive one. If we can't straighten out microsoft for them\, we can at least give them high quality tools.

Advocacy is a good thing\, right?

[p5pRT]

From @samtregar

On Sun\, 2 Apr 2000\, David Grove wrote​:

I'm not talking about corporate interference or selling out like [...] is trying to do with perl.

Is that "[...]" crap supposed to be more polite than outright slander?

Old marketing surveys from 96-97 showed that companies were moving from Novell and Unix to NT because they thought they had no choice. Since that time\, that choice has been a painful and expensive one. If we can't straighten out microsoft for them\, we can at least give them high quality tools.

Bizarre. You have a truely strange view of Perl's place in the universe. How exactly is Perl supposed to help people trying to use NT in place of Unix? Why should we want to?

Advocacy is a good thing\, right?

False. Advocacy *can* be a good thing. Advocacy can also be a very bad thing. It all depends on what you're advocating and how.

I imagine you can guess what a large portion of perl5-porters thinks of your kind of "advocacy" by now.

-sam