FAQ

On Sat 14 Feb 2004 03:33, Dave Mitchell wrote:
Some notes:

Pumpkings: Currently it requires bison 1.875 to rebuild perly.*, although
this restriction has more to do with a lack of testing on my part of other
versions, rather than any intrinsic defect. Probably 1.35 onwards will
work.
Looks fine.

A Question:
Will this work backwards to 5.005.x?
If so, I guess I can remove all probing for byacc from Configure

--
H.Merijn Brand Amsterdam Perl Mongers (http://amsterdam.pm.org/)
using perl-5.6.1, 5.8.0, & 5.9.x, and 806 on HP-UX 10.20 & 11.00, 11i,
AIX 4.3, SuSE 8.2, and Win2k. http://www.cmve.net/~merijn/
http://archives.develooper.com/daily-build@perl.org/ perl-qa@perl.org
send smoke reports to: smokers-reports@perl.org, QA: http://qa.perl.org

Search Discussions

  • Rafael Garcia-Suarez at Feb 14, 2004 at 11:03 am

    "H.Merijn Brand" wrote:
    On Sat 14 Feb 2004 03:33, Dave Mitchell wrote:
    Some notes:

    Pumpkings: Currently it requires bison 1.875 to rebuild perly.*, although
    this restriction has more to do with a lack of testing on my part of other
    versions, rather than any intrinsic defect. Probably 1.35 onwards will
    work.
    Looks fine.

    A Question:
    Will this work backwards to 5.005.x?
    If so, I guess I can remove all probing for byacc from Configure
    No, because byacc is used to build a2p. (Until now :)
  • Dave Mitchell at Feb 15, 2004 at 5:31 pm

    On Sat, Feb 14, 2004 at 12:02:14PM +0100, Rafael Garcia-Suarez wrote:
    "H.Merijn Brand" wrote:
    On Sat 14 Feb 2004 03:33, Dave Mitchell wrote:
    Some notes:

    Pumpkings: Currently it requires bison 1.875 to rebuild perly.*, although
    this restriction has more to do with a lack of testing on my part of other
    versions, rather than any intrinsic defect. Probably 1.35 onwards will
    work.
    Looks fine.

    A Question:
    Will this work backwards to 5.005.x?
    If so, I guess I can remove all probing for byacc from Configure
    No, because byacc is used to build a2p. (Until now :)
    Is that some kind of hint? ;-)

    Anyway, I was envisaging that the patch might go into 5.8.x, but it's
    definitely too scary for anything earlier. All the old byacc parser
    bolierplate code plus all its hacks to make it work under Perl have been
    thrown away and replaced with a whole new set of boilerplate code from
    bison plus a new set of hacks (not as scary hacks as the old one, as the
    new bison with the %pure_parser option doesn't use global vars any more).

    Dave.

    --
    Nothing ventured, nothing lost.
  • Nicholas Clark at Feb 14, 2004 at 5:22 pm

    On Sat, Feb 14, 2004 at 01:46:06PM +0000, Dave Mitchell wrote:

    Anyway, I was envisaging that the patch might go into 5.8.x, but it's
    definitely too scary for anything earlier. All the old byacc parser
    Depends how scarey it looks. How close is its output to the current *.c files,
    given the same input?
    Dave.

    --
    Nothing ventured, nothing lost.
    Oooh. A mission statement for maint. :-)

    Nicholas Clark
  • Dave Mitchell at Feb 15, 2004 at 5:32 pm

    On Sat, Feb 14, 2004 at 05:22:46PM +0000, Nicholas Clark wrote:
    On Sat, Feb 14, 2004 at 01:46:06PM +0000, Dave Mitchell wrote:

    Anyway, I was envisaging that the patch might go into 5.8.x, but it's
    definitely too scary for anything earlier. All the old byacc parser
    Depends how scarey it looks. How close is its output to the current *.c files,
    given the same input?
    Well, its like not close at all really. The boilerplate code is totally
    different, and the tables generally have different names and totally
    different content. Also, error recovery seems to be better.
    Nothing ventured, nothing lost.
    Oooh. A mission statement for maint. :-)
    Hmm, perhaps that means I've got the mental aptitude to be a maint pumpkin?
    On the other hand, considering what a scary thing I've just done to
    perly.*, I think I'm automatically disqualified :-)

    PS. having spent all of Valentine's day sorting out parser issues, I
    suspect that perhaps I need to get a life ;-)

    Dave.

    --
    A power surge on the Bridge is rapidly and correctly diagnosed as a faulty
    capacitor by the highly-trained and competent engineering staff.
    -- Things That Never Happen in "Star Trek" #9
  • Nicholas Clark at Feb 16, 2004 at 11:58 pm

    On Sun, Feb 15, 2004 at 01:17:27AM +0000, Dave Mitchell wrote:
    On Sat, Feb 14, 2004 at 05:22:46PM +0000, Nicholas Clark wrote:
    On Sat, Feb 14, 2004 at 01:46:06PM +0000, Dave Mitchell wrote:

    Nothing ventured, nothing lost.
    Oooh. A mission statement for maint. :-)
    Hmm, perhaps that means I've got the mental aptitude to be a maint pumpkin?
    On the other hand, considering what a scary thing I've just done to
    perly.*, I think I'm automatically disqualified :-)
    Yes, but after what I did just before 5.8.2, I should be disqualified.
    PS. having spent all of Valentine's day sorting out parser issues, I
    suspect that perhaps I need to get a life ;-)
    Yes, but I spent a lot of it converting Ton Hospel's perl gunzip prototype
    into a pure perl emulation of Compress::Zlib's inflate routine.

    I got to write some code using this lovely language "perl". Have you heard
    of it? I think that I should do this more often, as it was fun.

    Nicholas Clark

    PS "perl" must be OK, as it lets me exit a subroutine with last, and supports
    goto :-)
  • Dave Mitchell at Feb 17, 2004 at 1:38 pm

    On Mon, Feb 16, 2004 at 11:57:33PM +0000, Nicholas Clark wrote:
    I got to write some code using this lovely language "perl". Have you heard
    of it? I think that I should do this more often, as it was fun.
    Is that some sort of VB variant then?

    --
    "Emacs isn't a bad OS once you get used to it.
    It just lacks a decent editor."
  • Nicholas Clark at Feb 27, 2004 at 11:40 am

    On Sun, Feb 15, 2004 at 01:17:27AM +0000, Dave Mitchell wrote:
    On Sat, Feb 14, 2004 at 05:22:46PM +0000, Nicholas Clark wrote:
    On Sat, Feb 14, 2004 at 01:46:06PM +0000, Dave Mitchell wrote:

    Anyway, I was envisaging that the patch might go into 5.8.x, but it's
    definitely too scary for anything earlier. All the old byacc parser
    Depends how scarey it looks. How close is its output to the current *.c files,
    given the same input?
    Well, its like not close at all really. The boilerplate code is totally
    different, and the tables generally have different names and totally
    different content. Also, error recovery seems to be better.
    But thinking about merging it, I ask myself "Is it binary compatible?"
    and I don't think I can answer this myself.

    Nicholas Clark
  • Dave Mitchell at Feb 27, 2004 at 9:53 pm

    On Fri, Feb 27, 2004 at 11:39:32AM +0000, Nicholas Clark wrote:
    On Sun, Feb 15, 2004 at 01:17:27AM +0000, Dave Mitchell wrote:
    On Sat, Feb 14, 2004 at 05:22:46PM +0000, Nicholas Clark wrote:
    On Sat, Feb 14, 2004 at 01:46:06PM +0000, Dave Mitchell wrote:

    Anyway, I was envisaging that the patch might go into 5.8.x, but it's
    definitely too scary for anything earlier. All the old byacc parser
    Depends how scarey it looks. How close is its output to the current *.c files,
    given the same input?
    Well, its like not close at all really. The boilerplate code is totally
    different, and the tables generally have different names and totally
    different content. Also, error recovery seems to be better.
    But thinking about merging it, I ask myself "Is it binary compatible?"
    and I don't think I can answer this myself.
    Well, I *think* it's binary compatible. It makes a whole bunch of Perl
    vars obsolete (like PL_yydebug), but in the main patch (22302) I filled
    thinggyvar.h with placeholders, and in a seaprate patch for 5.9.x only
    (22303) I removed them altogether. Of course if someone's written
    some XS code that directly calls Perl_yylex or uses PL_yydebug, then it
    will break, but I can't really see that happening.

    So in short, yes, but don't ask me to stake my mother's life on it!

    --
    "The GPL violates the U.S. Constitution, together with copyright,
    antitrust and export control laws"
    -- SCO smoking crack again.
  • Dave Mitchell at Feb 27, 2004 at 11:16 pm

    On Fri, Feb 27, 2004 at 11:39:32AM +0000, Nicholas Clark wrote:
    But thinking about merging it, I ask myself "Is it binary compatible?"
    and I don't think I can answer this myself.
    Another consideration that has just occured to me (and relevant to 5.9.1
    too), is that I haven't had any feedback yet (good or bad) from EDCDIC
    users; since I wrote the code 'blind' to handle that side of things, it
    would be nice to know that it works before it gets into the wild.

    Dave.

    --
    In the 70's we wore flares because we didn't know any better.
    What possible excuse does the current generation have?
  • Nicholas Clark at Feb 28, 2004 at 3:15 pm

    On Fri, Feb 27, 2004 at 11:17:05PM +0000, Dave Mitchell wrote:
    On Fri, Feb 27, 2004 at 11:39:32AM +0000, Nicholas Clark wrote:
    But thinking about merging it, I ask myself "Is it binary compatible?"
    and I don't think I can answer this myself.
    Another consideration that has just occured to me (and relevant to 5.9.1
    too), is that I haven't had any feedback yet (good or bad) from EDCDIC
    users; since I wrote the code 'blind' to handle that side of things, it
    would be nice to know that it works before it gets into the wild.
    OK. This makes me wonder whether I wait until a blead snapshot (or the 5.9.1
    release) on the assumption that it will be easier to get someone to test
    EDCDIC from a tarball than by rsyncing.

    I'm going to did through Jarkko's contact list to see who might be able
    to test.

    Nicholas Clark
  • Pprymmer at Feb 28, 2004 at 10:45 pm
    putting together a tarball and announcing it on perl-mvs@perl.org might
    elicit some
    response (sorry but I personally no longer have access to z/OS anymore).

    Peter Prymmer




    Nicholas Clark
    <nick@ccl4.org>
    Sent by: Nicholas To
    Clark Dave Mitchell
    <nick@flirble.org <davem@fdisolutions.com>
    cc
    Rafael Garcia-Suarez
    <rgarciasuarez@free.fr>, "H.Merijn
    02/28/2004 10:15 Brand" <h.m.brand@hccnet.nl>,
    AM perl5-build@perl.org, Leon Brocard
    <acme@astray.com>
    Subject
    Re: switch from byacc to bison









    On Fri, Feb 27, 2004 at 11:17:05PM +0000, Dave Mitchell wrote:
    On Fri, Feb 27, 2004 at 11:39:32AM +0000, Nicholas Clark wrote:
    But thinking about merging it, I ask myself "Is it binary compatible?"
    and I don't think I can answer this myself.
    Another consideration that has just occured to me (and relevant to 5.9.1
    too), is that I haven't had any feedback yet (good or bad) from EDCDIC
    users; since I wrote the code 'blind' to handle that side of things, it
    would be nice to know that it works before it gets into the wild.
    OK. This makes me wonder whether I wait until a blead snapshot (or the
    5.9.1
    release) on the assumption that it will be easier to get someone to test
    EDCDIC from a tarball than by rsyncing.

    I'm going to did through Jarkko's contact list to see who might be able
    to test.

    Nicholas Clark

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupperl5-build @
categoriesperl
postedFeb 14, '04 at 9:20a
activeFeb 28, '04 at 10:45p
posts12
users5
websiteperl.org

People

Translate

site design / logo © 2019 Grokbase