[31783] in Perl-Users-Digest
Perl-Users Digest, Issue: 3046 Volume: 11
daemon@ATHENA.MIT.EDU (Perl-Users Digest)
Tue Jul 27 06:09:21 2010
Date: Tue, 27 Jul 2010 03:09:06 -0700 (PDT)
From: Perl-Users Digest <Perl-Users-Request@ruby.OCE.ORST.EDU>
To: Perl-Users@ruby.OCE.ORST.EDU (Perl-Users Digest)
Perl-Users Digest Tue, 27 Jul 2010 Volume: 11 Number: 3046
Today's topics:
4D hash iteration <a@a.com>
Re: 4D hash iteration <uri@StemSystems.com>
Re: Literals inside regular expression <nospam-abuse@ilyaz.org>
Re: Need help/advice to improve script sln@netherlands.com
Posting Guidelines for comp.lang.perl.misc ($Revision: tadmc@seesig.invalid
Re: Speed of reading some MB of data using qx(...) <nospam-abuse@ilyaz.org>
Re: Speed of reading some MB of data using qx(...) <ben@morrow.me.uk>
Re: Speed of reading some MB of data using qx(...) <w.c.humann@arcor.de>
Re: Speed of reading some MB of data using qx(...) <uri@StemSystems.com>
Re: Speed of reading some MB of data using qx(...) <w.c.humann@arcor.de>
Digest Administrivia (Last modified: 6 Apr 01) (Perl-Users-Digest Admin)
----------------------------------------------------------------------
Date: Tue, 27 Jul 2010 14:16:33 +0700
From: Michael <a@a.com>
Subject: 4D hash iteration
Message-Id: <i2m14b$2aem$1@adenine.netfront.net>
I need a construction like this:
foreach $a(...)
{
foreach $b(...)
{
foreach $c(...)
{
foreach $d(...)
{
$myval=$hash{$a}{$b}{$c}{$d};
...
}
}
}
}
How to do this? What should be in braces (...)?
--- news://freenews.netfront.net/ - complaints: news@netfront.net ---
------------------------------
Date: Tue, 27 Jul 2010 03:27:24 -0400
From: "Uri Guttman" <uri@StemSystems.com>
Subject: Re: 4D hash iteration
Message-Id: <87iq41l6zn.fsf@quad.sysarch.com>
>>>>> "M" == Michael <a@a.com> writes:
M> I need a construction like this:
M> foreach $a(...)
M> {
M> foreach $b(...)
M> {
M> foreach $c(...)
M> {
M> foreach $d(...)
M> {
M> $myval=$hash{$a}{$b}{$c}{$d};
M> ...
M> }
M> }
M> }
M> }
M> How to do this? What should be in braces (...)?
read perldoc perldsc, perldoc perlreftut for how to do this. at each
loop you need to get the keys of the next lower level or get the hash
reference of the next lower level. those docs will help you figure out
what to do.
uri
--
Uri Guttman ------ uri@stemsystems.com -------- http://www.sysarch.com --
----- Perl Code Review , Architecture, Development, Training, Support ------
--------- Gourmet Hot Cocoa Mix ---- http://bestfriendscocoa.com ---------
------------------------------
Date: Mon, 26 Jul 2010 21:54:03 +0000 (UTC)
From: Ilya Zakharevich <nospam-abuse@ilyaz.org>
Subject: Re: Literals inside regular expression
Message-Id: <slrni4s0vr.22i.nospam-abuse@powdermilk.math.berkeley.edu>
On 2010-07-26, Ben Morrow <ben@morrow.me.uk> wrote:
>> What is the moral? I do not know! m{\(\([^(]*\)\)} is not better,
>> right? Fontification by CPerl helps a little bit, of course, but not
>> much.
>
> Yes (well, Vim rather than CPerl, of course... :) ).
Hmm... Can Vim highlight REx according to function, as CPerl can (at
least up to Emacs v21)?
>> Lisp has the notion of "escaping out of quoting"; so it would look
>> something like
>>
>> m{ (( (?` [^(]* ) )) }xq
>>
>> assuming //q means /\Q/, except that the part inside (?` ) is not
>> quoted...
>
> I like the idea, but not the syntax. It's a little too much like SGML
> CDATA for me: 'everything is literal *except* this special magic
> sequence you've forgotten about...'.
So do not use it. ;-) The actual problem is, of course, whether "you
can recognize it when you see it"... And the fact that ` "unquotes" is
kinda mneumonic...
> I think if I seriously wanted an
> improvement on \Q\E it would look like
>
> /\q{((} [^(]* \q{))}/x
>
> with the character following \q being a delimiter (matched or not)
> following the same rules as for m// &c.
The problem would be the interaction of \q with //x. Essentially,
there must be two versions: one for "//x is ignored within this" and
one for "whitespace is ignored inside this". Something like \q and
\Q... But \Q is already occupied... \q and \qq ?
Yours,
Ilya
------------------------------
Date: Mon, 26 Jul 2010 15:06:00 -0700
From: sln@netherlands.com
Subject: Re: Need help/advice to improve script
Message-Id: <it0s46pmo4a7hd8cql2s8lgf32fdpfrg7l@4ax.com>
On Mon, 26 Jul 2010 00:54:34 -0700 (PDT), "sopan.shewale@gmail.com" <sopan.shewale@gmail.com> wrote:
>On Jul 23, 8:33 pm, "Peter J. Holzer" <hjp-usen...@hjp.at> wrote:
>> On 2010-07-22 18:22, s...@netherlands.com <s...@netherlands.com> wrote:
>>
>> > $line =~ /^
>> > ([^:]*)
>> > : (?<pass> [^:]*)
>> > : (?<emails> [^:]*)
>> > :? (?<flag> [^:]*)
>> > :? (?<pass_change> [^:]*)
>> > :? (?<flag_change> .*)
>> > $/x
>> > and $data->{$1} = {%+};
>>
>> That's nice. I guess one of these days I should try to get rid of 5.8.x
>> so that I can finally start to use 5.10 features ...
>>
>> hp
>
>[sopan] - this is interesting feature and not sure if it will work on
>old versions of Perl.
The original point was on fixing the regular expression to provide default ''
instead of undef(ined). Split is nice, its simple and mostly dumb. It uses
regular expressions as well, but with the focus on a partition, not the data.
Ok, here you go ... compatible with v5.8. It doesen't look too different from
the named capture buffers does it? It keeps a clear, readable, and modifyable
form.
The REAL advantage of named capture buffers is that there is no need to keep
track of sequentially numbered capture buffers. You can modify the regular
expression, insert/extract sub-expressions without the need to worry about
modifying the code body that uses the result.
-sln
-------------------
use strict;
use warnings;
use Data::Dumper;
my $data = {};
while (defined (my $line = <DATA>)) {
chomp $line;
$line =~ /^([^:]*):([^:]*):([^:]*):?([^:]*):?([^:]*):?(.*)$/
and $data->{$1} = {
pass =>$2,
emails =>$3,
flag =>$4,
pass_change =>$5,
flag_change =>$6
};
}
print Dumper($data);
__DATA__
AllPresent:hjliEO35kCgwI:all@example.com:1:23232:24324
LastMissing:CyL92g3OKi.jM:lastmissing@example.com:1:2323
FlagZero:CyL92g3OKi.jM:flagzero@example.com:0:23232
OnlyFlag:ZuqpLZ7AxHBvw:onlyflag@example.com:0
RestMissing:ZuqpLZ7AxHBvw:missing@example.com
------------------------------
Date: Tue, 27 Jul 2010 02:12:12 -0500
From: tadmc@seesig.invalid
Subject: Posting Guidelines for comp.lang.perl.misc ($Revision: 1.9 $)
Message-Id: <mYydnQ3b19VRG9PRnZ2dnUVZ_jydnZ2d@giganews.com>
Outline
Before posting to comp.lang.perl.misc
Must
- Check the Perl Frequently Asked Questions (FAQ)
- Check the other standard Perl docs (*.pod)
Really Really Should
- Lurk for a while before posting
- Search a Usenet archive
If You Like
- Check Other Resources
Posting to comp.lang.perl.misc
Is there a better place to ask your question?
- Question should be about Perl, not about the application area
How to participate (post) in the clpmisc community
- Carefully choose the contents of your Subject header
- Use an effective followup style
- Speak Perl rather than English, when possible
- Ask perl to help you
- Do not re-type Perl code
- Provide enough information
- Do not provide too much information
- Do not post binaries, HTML, or MIME
Social faux pas to avoid
- Asking a Frequently Asked Question
- Asking a question easily answered by a cursory doc search
- Asking for emailed answers
- Beware of saying "doesn't work"
- Sending a "stealth" Cc copy
Be extra cautious when you get upset
- Count to ten before composing a followup when you are upset
- Count to ten after composing and before posting when you are upset
-----------------------------------------------------------------
Posting Guidelines for comp.lang.perl.misc ($Revision: 1.9 $)
This newsgroup, commonly called clpmisc, is a technical newsgroup
intended to be used for discussion of Perl related issues (except job
postings), whether it be comments or questions.
As you would expect, clpmisc discussions are usually very technical in
nature and there are conventions for conduct in technical newsgroups
going somewhat beyond those in non-technical newsgroups.
The article at:
http://www.catb.org/~esr/faqs/smart-questions.html
describes how to get answers from technical people in general.
This article describes things that you should, and should not, do to
increase your chances of getting an answer to your Perl question. It is
available in POD, HTML and plain text formats at:
http://www.rehabitation.com/clpmisc.shtml
For more information about netiquette in general, see the "Netiquette
Guidelines" at:
http://andrew2.andrew.cmu.edu/rfc/rfc1855.html
A note to newsgroup "regulars":
Do not use these guidelines as a "license to flame" or other
meanness. It is possible that a poster is unaware of things
discussed here. Give them the benefit of the doubt, and just
help them learn how to post, rather than assume that they do
know and are being the "bad kind" of Lazy.
A note about technical terms used here:
In this document, we use words like "must" and "should" as
they're used in technical conversation (such as you will
encounter in this newsgroup). When we say that you *must* do
something, we mean that if you don't do that something, then
it's unlikely that you will benefit much from this group.
We're not bossing you around; we're making the point without
lots of words.
Do *NOT* send email to the maintainer of these guidelines. It will be
discarded unread. The guidelines belong to the newsgroup so all
discussion should appear in the newsgroup. I am just the secretary that
writes down the consensus of the group.
Before posting to comp.lang.perl.misc
Must
This section describes things that you *must* do before posting to
clpmisc, in order to maximize your chances of getting meaningful replies
to your inquiry and to avoid getting flamed for being lazy and trying to
have others do your work.
The perl distribution includes documentation that is copied to your hard
drive when you install perl. Also installed is a program for looking
things up in that (and other) documentation named 'perldoc'.
You should either find out where the docs got installed on your system,
or use perldoc to find them for you. Type "perldoc perldoc" to learn how
to use perldoc itself. Type "perldoc perl" to start reading Perl's
standard documentation.
Check the Perl Frequently Asked Questions (FAQ)
Checking the FAQ before posting is required in Big 8 newsgroups in
general, there is nothing clpmisc-specific about this requirement.
You are expected to do this in nearly all newsgroups.
You can use the "-q" switch with perldoc to do a word search of the
questions in the Perl FAQs.
Check the other standard Perl docs (*.pod)
The perl distribution comes with much more documentation than is
available for most other newsgroups, so in clpmisc you should also
see if you can find an answer in the other (non-FAQ) standard docs
before posting.
It is *not* required, or even expected, that you actually *read* all of
Perl's standard docs, only that you spend a few minutes searching them
before posting.
Try doing a word-search in the standard docs for some words/phrases
taken from your problem statement or from your very carefully worded
"Subject:" header.
Really Really Should
This section describes things that you *really should* do before posting
to clpmisc.
Lurk for a while before posting
This is very important and expected in all newsgroups. Lurking means
to monitor a newsgroup for a period to become familiar with local
customs. Each newsgroup has specific customs and rituals. Knowing
these before you participate will help avoid embarrassing social
situations. Consider yourself to be a foreigner at first!
Search a Usenet archive
There are tens of thousands of Perl programmers. It is very likely
that your question has already been asked (and answered). See if you
can find where it has already been answered.
One such searchable archive is:
http://groups.google.com/advanced_search
If You Like
This section describes things that you *can* do before posting to
clpmisc.
Check Other Resources
You may want to check in books or on web sites to see if you can
find the answer to your question.
But you need to consider the source of such information: there are a
lot of very poor Perl books and web sites, and several good ones
too, of course.
Posting to comp.lang.perl.misc
There can be 200 messages in clpmisc in a single day. Nobody is going to
read every article. They must decide somehow which articles they are
going to read, and which they will skip.
Your post is in competition with 199 other posts. You need to "win"
before a person who can help you will even read your question.
These sections describe how you can help keep your article from being
one of the "skipped" ones.
Is there a better place to ask your question?
Question should be about Perl, not about the application area
It can be difficult to separate out where your problem really is,
but you should make a conscious effort to post to the most
applicable newsgroup. That is, after all, where you are the most
likely to find the people who know how to answer your question.
Being able to "partition" a problem is an essential skill for
effectively troubleshooting programming problems. If you don't get
that right, you end up looking for answers in the wrong places.
It should be understood that you may not know that the root of your
problem is not Perl-related (the two most frequent ones are CGI and
Operating System related), so off-topic postings will happen from
time to time. Be gracious when someone helps you find a better place
to ask your question by pointing you to a more applicable newsgroup.
How to participate (post) in the clpmisc community
Carefully choose the contents of your Subject header
You have 40 precious characters of Subject to win out and be one of
the posts that gets read. Don't waste them. Take care while
composing them, they are the key that opens the door to getting an
answer.
Spend them indicating what aspect of Perl others will find if they
should decide to read your article.
Do not spend them indicating "experience level" (guru, newbie...).
Do not spend them pleading (please read, urgent, help!...).
Do not spend them on non-Subjects (Perl question, one-word
Subject...)
For more information on choosing a Subject see "Choosing Good
Subject Lines":
http://www.cpan.org/authors/id/D/DM/DMR/subjects.post
Part of the beauty of newsgroup dynamics, is that you can contribute
to the community with your very first post! If your choice of
Subject leads a fellow Perler to find the thread you are starting,
then even asking a question helps us all.
Use an effective followup style
When composing a followup, quote only enough text to establish the
context for the comments that you will add. Always indicate who
wrote the quoted material. Never quote an entire article. Never
quote a .signature (unless that is what you are commenting on).
Intersperse your comments *following* each section of quoted text to
which they relate. Unappreciated followup styles are referred to as
"top-posting", "Jeopardy" (because the answer comes before the
question), or "TOFU" (Text Over, Fullquote Under).
Reversing the chronology of the dialog makes it much harder to
understand (some folks won't even read it if written in that style).
For more information on quoting style, see:
http://web.presby.edu/~nnqadmin/nnq/nquote.html
Speak Perl rather than English, when possible
Perl is much more precise than natural language. Saying it in Perl
instead will avoid misunderstanding your question or problem.
Do not say: I have variable with "foo\tbar" in it.
Instead say: I have $var = "foo\tbar", or I have $var = 'foo\tbar',
or I have $var = <DATA> (and show the data line).
Ask perl to help you
You can ask perl itself to help you find common programming mistakes
by doing two things: enable warnings (perldoc warnings) and enable
"strict"ures (perldoc strict).
You should not bother the hundreds/thousands of readers of the
newsgroup without first seeing if a machine can help you find your
problem. It is demeaning to be asked to do the work of a machine. It
will annoy the readers of your article.
You can look up any of the messages that perl might issue to find
out what the message means and how to resolve the potential mistake
(perldoc perldiag). If you would like perl to look them up for you,
you can put "use diagnostics;" near the top of your program.
Do not re-type Perl code
Use copy/paste or your editor's "import" function rather than
attempting to type in your code. If you make a typo you will get
followups about your typos instead of about the question you are
trying to get answered.
Provide enough information
If you do the things in this item, you will have an Extremely Good
chance of getting people to try and help you with your problem!
These features are a really big bonus toward your question winning
out over all of the other posts that you are competing with.
First make a short (less than 20-30 lines) and *complete* program
that illustrates the problem you are having. People should be able
to run your program by copy/pasting the code from your article. (You
will find that doing this step very often reveals your problem
directly. Leading to an answer much more quickly and reliably than
posting to Usenet.)
Describe *precisely* the input to your program. Also provide example
input data for your program. If you need to show file input, use the
__DATA__ token (perldata.pod) to provide the file contents inside of
your Perl program.
Show the output (including the verbatim text of any messages) of
your program.
Describe how you want the output to be different from what you are
getting.
If you have no idea at all of how to code up your situation, be sure
to at least describe the 2 things that you *do* know: input and
desired output.
Do not provide too much information
Do not just post your entire program for debugging. Most especially
do not post someone *else's* entire program.
Do not post binaries, HTML, or MIME
clpmisc is a text only newsgroup. If you have images or binaries
that explain your question, put them in a publically accessible
place (like a Web server) and provide a pointer to that location. If
you include code, cut and paste it directly in the message body.
Don't attach anything to the message. Don't post vcards or HTML.
Many people (and even some Usenet servers) will automatically filter
out such messages. Many people will not be able to easily read your
post. Plain text is something everyone can read.
Social faux pas to avoid
The first two below are symptoms of lots of FAQ asking here in clpmisc.
It happens so often that folks will assume that it is happening yet
again. If you have looked but not found, or found but didn't understand
the docs, say so in your article.
Asking a Frequently Asked Question
It should be understood that you may have missed the applicable FAQ
when you checked, which is not a big deal. But if the Frequently
Asked Question is worded similar to your question, folks will assume
that you did not look at all. Don't become indignant at pointers to
the FAQ, particularly if it solves your problem.
Asking a question easily answered by a cursory doc search
If folks think you have not even tried the obvious step of reading
the docs applicable to your problem, they are likely to become
annoyed.
If you are flamed for not checking when you *did* check, then just
shrug it off (and take the answer that you got).
Asking for emailed answers
Emailed answers benefit one person. Posted answers benefit the
entire community. If folks can take the time to answer your
question, then you can take the time to go get the answer in the
same place where you asked the question.
It is OK to ask for a *copy* of the answer to be emailed, but many
will ignore such requests anyway. If you munge your address, you
should never expect (or ask) to get email in response to a Usenet
post.
Ask the question here, get the answer here (maybe).
Beware of saying "doesn't work"
This is a "red flag" phrase. If you find yourself writing that,
pause and see if you can't describe what is not working without
saying "doesn't work". That is, describe how it is not what you
want.
Sending a "stealth" Cc copy
A "stealth Cc" is when you both email and post a reply without
indicating *in the body* that you are doing so.
Be extra cautious when you get upset
Count to ten before composing a followup when you are upset
This is recommended in all Usenet newsgroups. Here in clpmisc, most
flaming sub-threads are not about any feature of Perl at all! They
are most often for what was seen as a breach of netiquette. If you
have lurked for a bit, then you will know what is expected and won't
make such posts in the first place.
But if you get upset, wait a while before writing your followup. I
recommend waiting at least 30 minutes.
Count to ten after composing and before posting when you are upset
After you have written your followup, wait *another* 30 minutes
before committing yourself by posting it. You cannot take it back
once it has been said.
AUTHOR
Tad McClellan and many others on the comp.lang.perl.misc newsgroup.
--
Tad McClellan
email: perl -le "print scalar reverse qq/moc.liamg\100cm.j.dat/"
The above message is a Usenet post.
I don't recall having given anyone permission to use it on a Web site.
------------------------------
Date: Mon, 26 Jul 2010 21:45:19 +0000 (UTC)
From: Ilya Zakharevich <nospam-abuse@ilyaz.org>
Subject: Re: Speed of reading some MB of data using qx(...)
Message-Id: <slrni4s0ff.22i.nospam-abuse@powdermilk.math.berkeley.edu>
On 2010-07-26, Peter J. Holzer <hjp-usenet2@hjp.at> wrote:
> On 2010-07-26 13:11, Ben Morrow <ben@morrow.me.uk> wrote:
>> If the issue simply turns out to be 'Microsoft don't know how to write a
>> decent malloc', there is very little p5p can do about it, of course. On
>> most platforms perl can, and often does, use its own malloc
>> implementation which is optimised for perl's use (lots of tiny
>> allocations and deallocations all the time). This isn't possible on
>> Win32 unless you make a custom build of perl that doesn't support the
>> fork emulation.
> Since the fork emulation works with Win32 malloc, I think it should be
> possible to write a custom malloc based on Win32 malloc (or the
> underlying API calls) which still works with the fork emulation but is
> faster. But it's probably not easy or somebody would have done it
> already (I don't pretend to understand either memory allocation or the
> fork emulation on windows).
"My" malloc() (one shipped with Perl) needs only 1 syscall implemented
on a particular architecture: get some pages from the system. So the
macro should be defined on the command line of $(CC) -c malloc.c.
Hope this helps,
Ilya
------------------------------
Date: Tue, 27 Jul 2010 01:15:28 +0100
From: Ben Morrow <ben@morrow.me.uk>
Subject: Re: Speed of reading some MB of data using qx(...)
Message-Id: <02l1i7-6lp1.ln1@osiris.mauzo.dyndns.org>
Quoth Ilya Zakharevich <nospam-abuse@ilyaz.org>:
> On 2010-07-26, Peter J. Holzer <hjp-usenet2@hjp.at> wrote:
> > On 2010-07-26 13:11, Ben Morrow <ben@morrow.me.uk> wrote:
> >> If the issue simply turns out to be 'Microsoft don't know how to write a
> >> decent malloc', there is very little p5p can do about it, of course. On
> >> most platforms perl can, and often does, use its own malloc
> >> implementation which is optimised for perl's use (lots of tiny
> >> allocations and deallocations all the time). This isn't possible on
> >> Win32 unless you make a custom build of perl that doesn't support the
> >> fork emulation.
>
> > Since the fork emulation works with Win32 malloc, I think it should be
> > possible to write a custom malloc based on Win32 malloc (or the
> > underlying API calls) which still works with the fork emulation but is
> > faster. But it's probably not easy or somebody would have done it
> > already (I don't pretend to understand either memory allocation or the
> > fork emulation on windows).
>
> "My" malloc() (one shipped with Perl) needs only 1 syscall implemented
> on a particular architecture: get some pages from the system. So the
> macro should be defined on the command line of $(CC) -c malloc.c.
There is an option in the Win32 makefile to do exactly that, with a note
that it doesn't work with USE_IMP_SYS. I believe this has something to
do with needing a separate malloc pool for each thread (pseudo-process),
but I don't know. I would have thought, on the face of it, that it
should be possible to simply replace malloc with mymalloc wherever it
ends up being called, but I don't pretend to understand anything in
perlhost.h.
Ben
------------------------------
Date: Mon, 26 Jul 2010 23:14:28 -0700 (PDT)
From: Wolfram Humann <w.c.humann@arcor.de>
Subject: Re: Speed of reading some MB of data using qx(...)
Message-Id: <e8a177ad-20f1-4c0f-894b-4aa29c7db827@f15g2000vbq.googlegroups.com>
I do understand how an unefficient malloc would explain the chunk-size
dependency in the code. What I do not understand is the dependency on
the initial size of the string that I append to. Does Perl re-allocate
the complete string (at least from time to time) when I keep appending
to it? Or could there be other effects involved besides malloc speed ?
(Ideally I would profile Strawberry Perl on the C level but I don't
know how to do that...)
Wolfram
------------------------------
Date: Tue, 27 Jul 2010 02:20:25 -0400
From: "Uri Guttman" <uri@StemSystems.com>
Subject: Re: Speed of reading some MB of data using qx(...)
Message-Id: <874oflmonq.fsf@quad.sysarch.com>
>>>>> "WH" == Wolfram Humann <w.c.humann@arcor.de> writes:
WH> I do understand how an unefficient malloc would explain the chunk-size
WH> dependency in the code. What I do not understand is the dependency on
WH> the initial size of the string that I append to. Does Perl re-allocate
WH> the complete string (at least from time to time) when I keep appending
WH> to it? Or could there be other effects involved besides malloc speed ?
WH> (Ideally I would profile Strawberry Perl on the C level but I don't
WH> know how to do that...)
perl does a realloc (or its equiv) on strings which grow too big. i
believe it does a doubling each time (same as for hash buckets). if so,
a large initial size will affect things as it will mean fewer malloc
calls as it grows. if malloc is the bottleneck, this will show up.
uri
--
Uri Guttman ------ uri@stemsystems.com -------- http://www.sysarch.com --
----- Perl Code Review , Architecture, Development, Training, Support ------
--------- Gourmet Hot Cocoa Mix ---- http://bestfriendscocoa.com ---------
------------------------------
Date: Tue, 27 Jul 2010 00:18:42 -0700 (PDT)
From: Wolfram Humann <w.c.humann@arcor.de>
Subject: Re: Speed of reading some MB of data using qx(...)
Message-Id: <97d1bf29-b5b9-4d69-8543-087c8776fa40@h25g2000vba.googlegroups.com>
On 27 Jul., 08:20, "Uri Guttman" <u...@StemSystems.com> wrote:
> >>>>> "WH" =3D=3D Wolfram Humann <w.c.hum...@arcor.de> writes:
>
> =A0 WH> I do understand how an unefficient malloc would explain the chunk=
-size
> =A0 WH> dependency in the code. What I do not understand is the dependenc=
y on
> =A0 WH> the initial size of the string that I append to. Does Perl re-all=
ocate
> =A0 WH> the complete string (at least from time to time) when I keep appe=
nding
> =A0 WH> to it? Or could there be other effects involved besides malloc sp=
eed ?
> =A0 WH> (Ideally I would profile Strawberry Perl on the C level but I don=
't
> =A0 WH> know how to do that...)
>
> perl does a realloc (or its equiv) on strings which grow too big. i
> believe it does a doubling each time (same as for hash buckets). if so,
> a large initial size will affect things as it will mean fewer malloc
> calls as it grows. if malloc is the bottleneck, this will show up.
Unfortunately, that makes thing just more mysterious (at least for my
understanding):
> 1E5 chars + 1E4 x 1E2 chars: 94.4 ms
> 1E6 chars + 1E4 x 1E2 chars: 319.9 ms
> 1E7 chars + 1E4 x 1E2 chars: 2710.4 ms
While appending 1E6 chars (=3D 1E4 chunks of 1E2 chars each) to an
initial string size of 1E5 might require 3-4 reallocs, appending 1E6
chars to an initial string size of 1E7 chars involves at most 1
realloc. And (even though not measured in my sample code above) the
from-scratch en-bloc allocation of a 2E7 chars string is reasonably
fast in Strawberry Perl.
Wolfram
------------------------------
Date: 6 Apr 2001 21:33:47 GMT (Last modified)
From: Perl-Users-Request@ruby.oce.orst.edu (Perl-Users-Digest Admin)
Subject: Digest Administrivia (Last modified: 6 Apr 01)
Message-Id: <null>
Administrivia:
To submit articles to comp.lang.perl.announce, send your article to
clpa@perl.com.
Back issues are available via anonymous ftp from
ftp://cil-www.oce.orst.edu/pub/perl/old-digests.
#For other requests pertaining to the digest, send mail to
#perl-users-request@ruby.oce.orst.edu. Do not waste your time or mine
#sending perl questions to the -request address, I don't have time to
#answer them even if I did know the answer.
------------------------------
End of Perl-Users Digest V11 Issue 3046
***************************************
