Date: Thu, 4 Mar 1999 00:49:03 -0600 (CST)
From: bradapp@enteract.com (Brad Appleton)
Subject: Re: cmp
To: walker@ncbi.nlm.nih.gov (D Roland Walker)
Cc: ppt@perl.com
Reply-to: bradapp@enteract.com
In-reply-to: <199903040425.WAA25154@agogic.cig.mot.com> from "D Roland Walker" a
t Mar 3, 99 11:16:27 pm
Delivery-Date: Wed Mar 3 23:48:49 1999
D Roland Walker writes:
> # It's nice to see that, speedwise, this beats the pants off
> # GNU and IRIX system cmp in many tasks. If compiled, this
> # would probably be faster in every case. Go Perl !
Glad to hear it. I assume that is why you chose to do it
the way you did instead of righting a ten-liner that invoked
File::Compare::cmp? Or was there another reason?
--
Brad Appleton <bradapp@enteract.com> | http://www.enteract.com/~bradapp/
"And miles to go before I sleep." | 3700+ WWW links on CS & Sw-Eng
Date: Thu, 4 Mar 1999 10:07:29 -0500
From: D Roland Walker <walker@ncbi.nlm.nih.gov>
Subject: Re: cmp
To: bradapp@enteract.com
Cc: ppt@perl.com
In-reply-to: <199903040651.BAA08872@po_box.cig.mot.com>; from Brad Appleton on T
hu, Mar 04, 1999 at 12:49:03AM -0600
Delivery-Date: Thu Mar 4 08:07:11 1999
[bradapp writes]
> D Roland Walker writes:
> > # It's nice to see that, speedwise, this beats the pants off
> > # GNU and IRIX system cmp in many tasks. If compiled, this
> > # would probably be faster in every case. Go Perl !
>
> Glad to hear it. I assume that is why you chose to do it
> the way you did instead of righting a ten-liner that invoked
> File::Compare::cmp? Or was there another reason?
Manily because File::Compare::cmp would really only support
cmp -s
and we need to be able support the plain default, not to mention
the other options and warnings. We could do with a cmp.simple
that just calls File::Compare::cmp, but since the *.simple are
also going to find value as teaching aids, I think it would be
even better to give another complete version of cmp that uses
more idiomatic and concise Perl to do a line-by-line compare.
One can easily imagine a
*.practical # like the cmp I submitted
*.simple # like a cmp that uses File::Compare
*.fancy # like a cmp that uses Mmap
*.pedagogy # like an idiomatic line-by-line cmp
for most of these tools. In fact, it would be cool to do them all.
R
Date: Thu, 4 Mar 1999 09:31:54 -0600 (CST)
From: bradapp@enteract.com (Brad Appleton)
Subject: Re: cmp
To: walker@ncbi.nlm.nih.gov (D Roland Walker)
Cc: ppt@perl.com
Reply-to: bradapp@enteract.com
In-reply-to: <199903041516.JAA04792@agogic.cig.mot.com> from "D Roland Walker" a
t Mar 4, 99 10:07:29 am
Delivery-Date: Thu Mar 4 08:31:44 1999
D Roland Walker writes:
> Mainly because File::Compare::cmp would really only support
>
> cmp -s
>
> and we need to be able support the plain default, not to mention
> the other options and warnings.
Okay, perhaps we could submit patches (or a replacement) to
File::Compare instead of having duplicated effort & maintenance.
> even better to give another complete version of cmp that uses
> more idiomatic and concise Perl to do a line-by-line compare.
Since you mention it ;-) Such a patch was just supplied to
File::Compare in the past couple weeks (in fact that was the thread on
P5P which spawned off Tom's PPT project reply). Jan Dubois added some
code so that if the (optional) third argument is a negative buffersize,
then it does a line by line text-compare, *and* if the third argument
is a subroutine-ref, then it calls the supplied sub to compare lines
(to perform any "approximate" rather than exact comparison). It would
be good if the good stuff that you've done here could be integrated with
that.
> *.fancy # like a cmp that uses Mmap
The above could probably be done by creating an IO::MMap (which doesn't
exist yet, and probably inst portable ;-) and passing it to
File::Compare.
> *.pedagogy # like an idiomatic line-by-line cmp
Line-by-line is good more than simply pedagogy. Jan's changes wouldn't
have been needed otherwise. The problem it solved here is that using a
binary, byte-per-byte (or block-per-block) comparison will consider
files with lines that end in "\r\n" different from those using only
"\n" even if they are the same in all other respects. To do something
like compare expected text output generated on one platform to actual
output generated on another platform, the line-compare using chomped
<lines> will have the desired result.
Cheers!
--
Brad Appleton <bradapp@enteract.com> | http://www.enteract.com/~bradapp/
"And miles to go before I sleep." | 3700+ WWW links on CS & Sw-Eng