[prev in list] [next in list] [prev in thread] [next in thread] 

List:       perl-mailbox
Subject:    Re: Mail::Box 2.022 make test failure in perl 5.8.0
From:       Greg Matheson <cid88201 () mail ! chinmin ! edu ! tw>
Date:       2002-09-10 13:26:25
[Download RAW message or body]

On Tue, 10 Sep 2002, Mark Overmeer wrote:

> The tests often expect specific sizes in the Content-Length fields,
> which depend on the line-terminator.  This should be solved on a
> test-for-test basis :(  That's why I am afraid that either I must find
> a cygwin machine and learn how it works, or find someone with too much
> time at hand and a lot of experience...

Well, cygwin is the Unix look-and-feel on a Windows machine.
One cygwin slogan: "Never leave Unix without it."

> > So I guess knowing the size of a message on Win32 depends on
> > knowing how many lines it has. Or perhaps you know, at least it won't be
> > any bigger than size(file).

> The rules are quite simple:
>   Content-Length(win32) = Content-Length(unix) + Lines;

> But that does not solve the problem of the failing tests.

I hear that most mailing software that could do things like use
Content-Length to parse an mbox folder into its messages has that
capability normally turned off. That's because it's not robust. A byte
missing here and there, and the mailbox becomes unreadable.

"From " line parsing on the other hand is less affected by local
errors.

And that's just Unix we're talking about. So using Content-Length
over different OS is even less robust.

However, it could still be used where a margin of error is
acceptable, or where you are interested in knowing a maximum
bound or something.

-- 
Greg Matheson            Failed to turn my reluctant students into autonomous
Chinmin College          language learners, but it did wonders for my personal
                         fitness program.
Taiwan Penpals Archive <URL: http://netcity.hinet.net/kurage>


[prev in list] [next in list] [prev in thread] [next in thread]