At 5:13 PM -0800 3/28/03, Michael G Schwern wrote:

>Show me your descrip.mms and failures.

descrip.mms and log of complete build and test cycle attached.  Note
that there are occasional Perl warnings that don't seem to cause test
failures, such as at 00compile.t #11.

The Perl crashes do not occur when running the tests individually --
only when running through test_harness.  I'll try to investigate this
further, though it's proving to be pretty elusive.

> > 1.) When forcing the extension to build itself with its own libraries,
>> the paths have to be relative, not absolute. Among other things, that
>> keeps the Perl command used for one-liners from getting too long.
>
>They must be absolute because things chdir around while building and
>testing. 

MM's has always chdir'd around and it's always used relative paths to do so.

>The Perl command hasn't gotten too long for me, are you seeing
>problems?

Yes, though not in the test suite.  But even if it works for you and
for me, it's going to break for someone with a long path name.
Compare the following two liines:

$ mcr perl_root:[000000]miniperl.exe "-I[.lib]"

$ mcr perl_root:[000000]miniperl.exe 
"-IDISK$USER100:[USERS.JOHNBROWN.EXTUTILS-MAKEMAKER-SNAP.lib]"

The first one is short, relocatable, and predictable.  The second
method consumes a huge portion of the allowed 255 character command
and it varies unpredictably based on the local disk and directory
names.

-- 
________________________________________
Craig A. Berry
mailto:[EMAIL PROTECTED]

"... getting out of a sonnet is much more
 difficult than getting in."
                 Brad Leithauser

Attachment: %make_mm.log
Description: application/applefile

Attachment: make_mm.log
Description: Binary data

Attachment: %descrip.mms
Description: application/applefile

Attachment: descrip.mms
Description: Binary data

Reply via email to