Re: [9fans] Re: Perl5 & kenji arisawa's perl question

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: <rog@vitanuova.com>

> luckily we already have vi.
> *phew*

well i hope the 'goto fonfon;' was kept -- its only saving grace.
at least you good laugh yourself silly at that one.

ever wondered why constructs like:

    10~

never worked?  that was because a bunch of stuff was implemented
by forcing stuff back into its input.

~ was implemented by peeking at the character the cursor was on,
changing its case and then forcing:

   r.l

back on input, where . is the character in question.  the
count gets thrown away in that ghastly main loop.

the 'l' would have been not added if you were at the
end of the line.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[rog]

> i would also object to [...] vi [...] being ported to plan 9.

luckily we already have vi.
*phew*

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: Steve Kilbane <steve@whitecrow.demon.co.uk>
> Having a prompt like:
> 
> hist ;
> 

good point.

programming the inputs.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Steve Kilbane]

> I don't believe it's easier to find a command in an acme window and
> then flaff around with mouse button 2 to execute it (trying not to
> accidentally execute half of the next line, too!)  than it is to hit
> the cursor-up key in old fashioned shells like tcsh.

I'd agree with that, but that only goes so far - the last few commands
are great, but then you start getting less return as you have to
linearly and manually search back through time.

Once you get into actually searching for things, then you're past the
simple up-arrow idea, and it's time to revisit. ye olde "=p pattern"
works well on a system that saves over time, and allows you to edit
the results before executing.

In addition to being "forever", the old rc history file was also
shareable across shells. With plan 9, it's shared across everything,
if you want. This may be good or bad, depending on your opinion
(and probably your mood).

> It's not a
> question of tyranny of the old ways, rather it's a question of ease of
> use, sometimes at the end of a hard day, and the number of muscles you
> need to use to perform a common operation.

I can't help but feel you can work around this, if you want. The great
thing about setting a prompt to whatever you want is that the prompt
can be your most commonly-executed command, if it helps. Having a
prompt like:

hist ;

could show your last ten commands immediately, ready for re-execution.

steve

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: <anothy@cosym.net>

> i will, however conceed that being able to search the history
> would be quite useful. Acme provides this automatically. i'm
> aware of the concern about making the button 2 menu in rio
> too long, but i still think i'd be in favor of "look".

i believe that was rob's argument against 'look'.  he wanted to
keep the button 2 menu short, which i agree with, but not in
this case.  guess that's a feature of open source;  i can add
it if i want it.

however i _would not_ add a forward/backward entry.  that
would make the menu far too long and i've always found
that you could zip around a window pretty quickly, just by
a forward search

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[bwc]

[-- Attachment #1: Type: text/plain, Size: 943 bytes --]

> I don't believe it's easier to find a command in an acme window and
> then flaff around with mouse button 2 to execute it (trying not to
> accidentally execute half of the next line, too!)  than it is to hit
> the cursor-up key in old fashioned shells like tcsh.

I keep the commands in a text file.  I don't use Win that much.

>   It's not a
> question of tyranny of the old ways, rather it's a question of ease of
> use, sometimes at the end of a hard day, and the number of muscles you
> need to use to perform a common operation.

I kind of like moving my arms.

> Also, the virtual roll of terminal paper has a splendid audit trail
> capability.  There's no chance of looking back in a Plan 9 window to
> see what you did previously because you've probably edited half the
> commands to make new ones!
> Dave Atkin

I suggest using ed(1) then, you even get an audit trail of your
editing. :-)

  Brantley Coile

[-- Attachment #2: Type: message/rfc822, Size: 2932 bytes --]

From: Dave Atkin <dla@vitanuova.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Re: Perl5 & kenji arisawa's perl question
Date: Thu, 9 Nov 2000 16:51:56 GMT
Message-ID: <973787527.26907.0.nnrp-01.d4f0e306@news.demon.co.uk>

As a relative newcomer to Plan 9 I find it a bit "mouse-centric".

I don't believe it's easier to find a command in an acme window and
then flaff around with mouse button 2 to execute it (trying not to
accidentally execute half of the next line, too!)  than it is to hit
the cursor-up key in old fashioned shells like tcsh.  It's not a
question of tyranny of the old ways, rather it's a question of ease of
use, sometimes at the end of a hard day, and the number of muscles you
need to use to perform a common operation.

Also, the virtual roll of terminal paper has a splendid audit trail
capability.  There's no chance of looking back in a Plan 9 window to
see what you did previously because you've probably edited half the
commands to make new ones!

Dave Atkin

> 
> Acme's use of mouse button 2 to execute a command addresses requirements
> of the history feature of shells.  This idea was imported from Wirth's Oberon.
> In Oberon you only have to type a command once and it live forever in a text
> window.  This is simular to the Acme guides.  You don't have to look
> through thousands of lines of output with some small number of commands all mixed.
> The Oberon and acme interfaces allow these functions for ANY program,
> not just the shell.  I don't miss the virtual roll of TI Silent 300 terminal paper.
> Look closely at Acme.  It takes some work to break free of the tyranny of the
> old ways.
> 
>  Brantley Coile
> 

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[anothy]

well, to answer your question first, no, rc has no command or
filename completion. moving on... if you found the "history"
provided by Plan 9 (i assume you mean rio) "useless" i suggest
you spend some more time getting used to it. being able to see
the output associated with a command a big win. the editing of
input and output is huge, too. they're certainly not the same,
but it's a far cry from "useless".

i will, however conceed that being able to search the history
would be quite useful. Acme provides this automatically. i'm
aware of the concern about making the button 2 menu in rio
too long, but i still think i'd be in favor of "look".
-α.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Dave Atkin]

As a relative newcomer to Plan 9 I find it a bit "mouse-centric".

I don't believe it's easier to find a command in an acme window and
then flaff around with mouse button 2 to execute it (trying not to
accidentally execute half of the next line, too!)  than it is to hit
the cursor-up key in old fashioned shells like tcsh.  It's not a
question of tyranny of the old ways, rather it's a question of ease of
use, sometimes at the end of a hard day, and the number of muscles you
need to use to perform a common operation.

Also, the virtual roll of terminal paper has a splendid audit trail
capability.  There's no chance of looking back in a Plan 9 window to
see what you did previously because you've probably edited half the
commands to make new ones!

Dave Atkin

> 
> Acme's use of mouse button 2 to execute a command addresses requirements
> of the history feature of shells.  This idea was imported from Wirth's Oberon.
> In Oberon you only have to type a command once and it live forever in a text
> window.  This is simular to the Acme guides.  You don't have to look
> through thousands of lines of output with some small number of commands all mixed.
> The Oberon and acme interfaces allow these functions for ANY program,
> not just the shell.  I don't miss the virtual roll of TI Silent 300 terminal paper.
> Look closely at Acme.  It takes some work to break free of the tyranny of the
> old ways.
> 
>  Brantley Coile
> 

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[erik quanstrom]

heck, i've got 7 years and 1/4 million lines of history. ;-)
i think that's the real utility of history. since rc lets
you do fairly complicated things on the command line,
sometimes the history is the only way to drege up the
code you wrote a couple of months^wdays ago.

erik

------------------

i have to say that although i get along with the plan 9 way quite
happily, i do have a sneaking regret for the rc history file that was
provided with unix rc. i never deleted it, which meant that if vaguely
remembered a slightly contorted ad hoc rc script that i'd typed some
months back, i could always find it (NFS file appending bugs
permitting...)

at the end of my last job the history file contained about 70,000
commands, if i remember rightly...

  rog.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[rog]

i have to say that although i get along with the plan 9 way quite
happily, i do have a sneaking regret for the rc history file that was
provided with unix rc. i never deleted it, which meant that if vaguely
remembered a slightly contorted ad hoc rc script that i'd typed some
months back, i could always find it (NFS file appending bugs
permitting...)

at the end of my last job the history file contained about 70,000
commands, if i remember rightly...

  rog.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

i believe that history was taken out of the plan 9 rc
'cos nobody used it.  rob is right.  all those horrible
gnu tools with their ridiculous options.  on unix i need
history, on plan 9 i don't.  the text is in the window.
just pick it up and send it.  if it's complex, write
a function or use 'mk'.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[bwc]

[-- Attachment #1: Type: text/plain, Size: 651 bytes --]

Acme's use of mouse button 2 to execute a command addresses requirements
of the history feature of shells.  This idea was imported from Wirth's Oberon.
In Oberon you only have to type a command once and it live forever in a text
window.  This is simular to the Acme guides.  You don't have to look
through thousands of lines of output with some small number of commands all mixed.
The Oberon and acme interfaces allow these functions for ANY program,
not just the shell.  I don't miss the virtual roll of TI Silent 300 terminal paper.
Look closely at Acme.  It takes some work to break free of the tyranny of the
old ways.

 Brantley Coile

[-- Attachment #2: Type: message/rfc822, Size: 2815 bytes --]

From: Greg Shubin <gshubin@sonic.net>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Re: Perl5 & kenji arisawa's perl question
Date: Thu, 9 Nov 2000 09:33:03 GMT
Message-ID: <3A09AE3B.8ABB4D07@sonic.net>

Russ Cox wrote:

> Dodging the C++ bomb shell (quiet, Boyd!), I think I
> would summarize my view of the feeling by saying that
> Plan 9 provides you with a new perspective on how
> to implement various things that in Unix are taken
> for granted as the ``only way to fly''.
>
> The bash features you cite are a great example: instead of
> endowing the shell with command completion, history
> using arrow keys, and emacs command editing, Plan 9 puts
> roughly equivalent capabilities in the hands of the window
> system: you can edit any text in any rio window, cut, paste,
> etc.  The feel is different, but now it applies to _all_ applications,
> not just the shell, and without requiring everything to link
> against the GNU readline library.

In my limited experience with Plan 9, I found the "history" just about
useless compared with bash. With bash, I can let the computer grep a few
thousand lines of history, and select whatever is appropriate. With the
Plan 9 cut and paste, I have to visually search, then move from the
keyboard to the mouse and back. Plus I only get a page or two of history.
Also, does Plan 9/rc have [command | filename] completion? I haven't
found it yet. (maybe I should RTFM?)

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Greg Shubin]

Russ Cox wrote:

> Dodging the C++ bomb shell (quiet, Boyd!), I think I
> would summarize my view of the feeling by saying that
> Plan 9 provides you with a new perspective on how
> to implement various things that in Unix are taken
> for granted as the ``only way to fly''.
>
> The bash features you cite are a great example: instead of
> endowing the shell with command completion, history
> using arrow keys, and emacs command editing, Plan 9 puts
> roughly equivalent capabilities in the hands of the window
> system: you can edit any text in any rio window, cut, paste,
> etc.  The feel is different, but now it applies to _all_ applications,
> not just the shell, and without requiring everything to link
> against the GNU readline library.

In my limited experience with Plan 9, I found the "history" just about
useless compared with bash. With bash, I can let the computer grep a few
thousand lines of history, and select whatever is appropriate. With the
Plan 9 cut and paste, I have to visually search, then move from the
keyboard to the mouse and back. Plus I only get a page or two of history.
Also, does Plan 9/rc have [command | filename] completion? I haven't
found it yet. (maybe I should RTFM?)

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Steve Kilbane]

> ok, i'll tell you one story:
> 
>    programs that compiled one day would not compile the next day
> 
> with this sort of reliability, would you choose it as a programming
> language -- and that's long before the bloat.

well, that's not particularly uncommon in the very early days of a
programming language. how many times has ken said they decided not
to change something in the early unix because they now had three
sites with installations? sun learned from that, and allowed java
to continue developing in the early days, leading to the same "problem".

where it becomes a problem is when the language *keeps* changing.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: andrey mirtchovski <aam396@mail.usask.ca>

> tell us the stories then... part of why people are attracted to p9 is the
> history that goes with it :)

c++ and plan 9 have nothing to do with each other.

i have a reliable source about c++, back in the '80s.

i wasn't there, but my source is/was reliable.

ok, i'll tell you one story:

   programs that compiled one day would not compile the next day

with this sort of reliability, would you choose it as a programming
language -- and that's long before the bloat.

those who forget their history are doomed to repeat it...

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[andrey mirtchovski]

On Wed, 8 Nov 2000, Boyd Roberts wrote:

> i won't even comment on c++.  i know the stories.
> 

tell us the stories then... part of why people are attracted to p9 is the
history that goes with it :)

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: Russ Cox <rsc@plan9.bell-labs.com>

> Dodging the C++ bomb shell (quiet, Boyd...

i won't even comment on c++.  i know the stories.

fine reply, russ.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: Greg Shubin <gshubin@sonic.net>
> 
> (Personally I like the features of bash (completion, history using arrow
> keys, emacs command editing, etc). )

bash...

you have not understood rc.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: <rog@vitanuova.com>
> > > luckily we already have vi.
> > > *phew*
> >
> > well i hope the 'goto fonfon;' was kept -- its only saving grace.
> 
> % cd /sys/src/cmd/vi
> % grep fonfon *.[ch]
> % 
> 
> i'm not sure what you're on about.
> :-)

i'm talking about the _original_ BSD source.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Russ Cox]

Dodging the C++ bomb shell (quiet, Boyd!), I think I 
would summarize my view of the feeling by saying that
Plan 9 provides you with a new perspective on how
to implement various things that in Unix are taken
for granted as the ``only way to fly''.  

The bash features you cite are a great example: instead of 
endowing the shell with command completion, history 
using arrow keys, and emacs command editing, Plan 9 puts
roughly equivalent capabilities in the hands of the window
system: you can edit any text in any rio window, cut, paste,
etc.  The feel is different, but now it applies to _all_ applications,
not just the shell, and without requiring everything to link
against the GNU readline library.

Similarly, the editors are of a different flavor: simpler, 
less to understand, but just as powerful when you do.

Plan 9 is neat because most things that are part of
the center of the system have been rethought and 
redesigned at least once more than Unix counterparts.

At the same time, no one is going to argue that Plan 9
is complete: there are plenty of things I want to be able
to do that I can't.  Reading news is one of them.  But if
someone were going to expend the effort to get news
running on Plan 9, I'd rather see a file system presentation
than just a port of trn.

Bringing in tools from Unix usually happens because they're
needed for some particular job: I have a perl binary because
I had to run one Perl script with some frequency when I was
working with a die-hard Perl fan on a project.  I have cvs
because I had to use it for a project last summer.  The 
TeX and Moscow ML ports happened for similar reasons.
That's why vncviewer exists too.

Those have been brought in because no one wanted to think
about redoing them, and they were needed.

Things like csh, sendmail, vi, and X we have replacements
for, and they're good examples of the Plan 9 approach to 
cleaner solutions for old problems.  If you were going to
use Plan 9 without using rc, upas, and rio, it really wouldn't
be much different from Unix.

Russ

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[David Gordon Hogan]

> At the risk of escalating language flame wars, what's your opinion of
> (wait a second, let me get into my bomb shelter) C++?

Nasssty language, precious.  It hurts poor Smeagol.  We hates it forever.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Greg Shubin]

Boyd Roberts wrote:

> From: George Michaelson <ggm@dstc.edu.au>
>
> > boyd@planete.net (Boyd Roberts) writes:
> >
> > >gotta be a better idea than the perl port :-)
> >
> > but would you of neccessity say the same thing of a python port? or
> > a tcl port? seems like it might be better to be explicit about
> > criticizing a language, not the idea of porting language(s) per se.
> >
>
> i'm criticising the language, not languages.  perl is an abortion;
> it is unreadable, the grammar is ghastly and it violates the 'tools'
> approach.
>
> i understand why sysadmins like it.  i just won't use it,
> although i have bug-fixed it.
>
> i would also object to csh, sendmail, vi, *rn, readnews, X
> and various other pieces of unix trash being ported to plan 9.
> it's plan 9, if you want unix, you know where to find it.

As a Plan 9 newbie, that's one of the things that I don't quite get. On
one hand it seems that the "official" position is Plan 9 contains
everything you need. But then I see everyone porting their favorite
tools to it. As you point out, pretty soon it will have as much bloat as
Un*x. Is there an official position on what should and shouldn't be
ported?

(Personally I like the features of bash (completion, history using arrow
keys, emacs command editing, etc). )

At the risk of escalating language flame wars, what's your opinion of
(wait a second, let me get into my bomb shelter) C++?

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[rog]

> > luckily we already have vi.
> > *phew*
>
> well i hope the 'goto fonfon;' was kept -- its only saving grace.

% cd /sys/src/cmd/vi
% grep fonfon *.[ch]
% 

i'm not sure what you're on about.
:-)

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[forsyth]

[-- Attachment #1: Type: text/plain, Size: 430 bytes --]

for those not seeing the joke: vi in plan 9 is a mips instruction interpreter.
emacs(1) is:

     EMACS(1)                                                 EMACS(1)

     NAME
          emacs - editor macros

     SYNOPSIS
          emacs [ options ]

     DESCRIPTION
          This page intentionally left blank.

     SOURCE
          MIT

     SEE ALSO
          sam(1), vi(1)

     BUGS
          Yes.


[-- Attachment #2: Type: message/rfc822, Size: 1465 bytes --]

To: cse.psu.edu!9fans
Subject: Re: [9fans] Re: Perl5 & kenji arisawa's perl question
Date: Wed, 8 Nov 2000 16:45:17 0000
Message-ID: <20001108154332.0BC19199E3@mail.cse.psu.edu>

> > luckily we already have vi.
> > *phew*
>
> well i hope the 'goto fonfon;' was kept -- its only saving grace.

% cd /sys/src/cmd/vi
% grep fonfon *.[ch]
% 

i'm not sure what you're on about.
:-)

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[anothy]

//luckily we already have vi.

good thing, too. but i think the man
page should include emacs(1) in the
SEE ALSO section. i much prefer that
port to plan 9. ☺
-α.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: George Michaelson <ggm@dstc.edu.au>

> boyd@planete.net (Boyd Roberts) writes:
>
> >gotta be a better idea than the perl port :-)
> 
> but would you of neccessity say the same thing of a python port? or
> a tcl port? seems like it might be better to be explicit about
> criticizing a language, not the idea of porting language(s) per se.
> 

i'm criticising the language, not languages.  perl is an abortion;
it is unreadable, the grammar is ghastly and it violates the 'tools'
approach.

i understand why sysadmins like it.  i just won't use it,
although i have bug-fixed it.

i would also object to csh, sendmail, vi, *rn, readnews, X
and various other pieces of unix trash being ported to plan 9.
it's plan 9, if you want unix, you know where to find it.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[George Michaelson]

boyd@planete.net (Boyd Roberts) writes:

>From: Douglas A. Gwyn <gwyn@arl.army.mil>
>> 
>> Hm, should I port my Adventure shell to rc?
>> 

>gotta be a better idea than the perl port :-)

but would you of neccessity say the same thing of a python port? or
a tcl port? seems like it might be better to be explicit about
criticizing a language, not the idea of porting language(s) per se.

(and I believe some bumf on inferno implied it would be a killer java
 platform)

-George

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[anothy]

//should I port my Adventure shell to rc?

oh, heck yeah! i'd rather have that
than perl and python combined!
-α.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

From: Douglas A. Gwyn <gwyn@arl.army.mil>
> 
> Hm, should I port my Adventure shell to rc?
> 

gotta be a better idea than the perl port :-)

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Douglas A. Gwyn]

Boyd Roberts wrote:
> From: <steve.simon@snellwilcox.com>
> > but what is tso(6)?
> well section 6 tells you it's a game.  it was a tso
> [time sharing option?] 'simulation'.

Hm, should I port my Adventure shell to rc?

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[forsyth]

[-- Attachment #1: Type: text/plain, Size: 245 bytes --]

i came across a copy of tso on tapes a few years ago and compiled
it (with changes for changed C and unix).  it was satisfying to see
a manly user interface once more.  i had more work to
explain the history to people to whom i showed it.


[-- Attachment #2: Type: message/rfc822, Size: 1968 bytes --]

To: <cse.psu.edu!9fans>
Subject: Re: [9fans] Re: Perl5 & kenji arisawa's perl question
Date: Mon, 6 Nov 2000 16:51:12 +0100
Message-ID: <023601c04809$63a46e80$0ab9c6d4@cybercable.fr>

----- Original Message ----- 
From: Scott Schwartz <schwartz@bio.cse.psu.edu>

> perl, python... anyone done tcl or es yet?

or tso(6) for that matter...

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Boyd Roberts]

----- Original Message ----- 
From: Scott Schwartz <schwartz@bio.cse.psu.edu>

> perl, python... anyone done tcl or es yet?

or tso(6) for that matter...

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Scott Schwartz]

perl, python... anyone done tcl or es yet?

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[nigel]

[-- Attachment #1: Type: text/plain, Size: 29 bytes --]

So where is it available?


[-- Attachment #2: Type: message/rfc822, Size: 2572 bytes --]

From: Jonathan Sergent <sergent@csociety.purdue.edu>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Re: Perl5 & kenji arisawa's perl question
Date: Mon, 06 Nov 2000 03:03:23 -0500
Message-ID: <200011060803.DAA12456@csociety.ecn.purdue.edu>

> 
> I also thank Nigel. Please give me Python 1.6.
> 

How's this:

$ ls -l python
--rwxr-xr-x M 3 sergent sergent 3193673 Nov  5 23:27 python
$ ./python
Python 2.0 (#3, Nov  5 2000, 23:27:32) [C] on plan91
Type "copyright", "credits" or "license" for more information.
>>> 

Check out that executable size...

Usually Python uses dynamic linking to load most of the modules.  If I get
up the courage, I will see if I can make GNU dld work (it does dynamic
module loading, supposedly portably, but I have not looked into it)
since Python already has support for it.

Out of the regression test suite, test_cmath, test_openpty, test_popen2,
and test_socket failed.  The first two caused python to suicide; the
second two generated error messages.  The test_socket error was a host
name resolution thing.  Does gethostbyname in APE work?  Does DNS work
with it?  Seems like someone mentioned this a while back.

Building Python was not that hard; I spent maybe two hours.

It's probably not that useful without the dynamic module loading, which
most Python programs I have seen use to load C code.  This would also
reduce the python executable size significantly...


--jss.

Re: [9fans] Re: Perl5 & kenji arisawa's perl question

[Jonathan Sergent]

> 
> I also thank Nigel. Please give me Python 1.6.
> 

How's this:

$ ls -l python
--rwxr-xr-x M 3 sergent sergent 3193673 Nov  5 23:27 python
$ ./python
Python 2.0 (#3, Nov  5 2000, 23:27:32) [C] on plan91
Type "copyright", "credits" or "license" for more information.
>>> 

Check out that executable size...

Usually Python uses dynamic linking to load most of the modules.  If I get
up the courage, I will see if I can make GNU dld work (it does dynamic
module loading, supposedly portably, but I have not looked into it)
since Python already has support for it.

Out of the regression test suite, test_cmath, test_openpty, test_popen2,
and test_socket failed.  The first two caused python to suicide; the
second two generated error messages.  The test_socket error was a host
name resolution thing.  Does gethostbyname in APE work?  Does DNS work
with it?  Seems like someone mentioned this a while back.

Building Python was not that hard; I spent maybe two hours.

It's probably not that useful without the dynamic module loading, which
most Python programs I have seen use to load C code.  This would also
reduce the python executable size significantly...


--jss.

[9fans] Re: Perl5 & kenji arisawa's perl question

[arisawa]

Hello 9fans,

Hi Andrey,
Your information is very helpful and I have installed perl5.004_05.
I will put perl5.004_05 on my ftp server (plan9.aichi-u.ac.jp) with
your comments. Thanks.

>redefinitions of 'FLT_MAX' apparently from float.h
FLT_MAX, FLT_MIN, DBL_MAX, DBL_MIN are defined in both ape/float.h
and ape/limit.h

I also thank Nigel. Please give me Python 1.6.

Kenji Arisawa
E-mail: arisawa@aichi-u.ac.jp