Re: [XML::XSLT] Use of Time::HiRes, OS dependant code

[Mike808 <mi...@ne...>]

"Mark A. Hershberger" wrote:
> That is, is this a case where we should educate the users to use
> URIs and not their native file scheme?  Or should we just do what is
> expedient and try to guess what the user means?"

Now I see where you're coming from. I vote for the former (educate
users about URIs).

I went back and looked at the specifics of the $OS usage and here's
what I found. We still need it.

When spitting out a document for a webserver, you must send two CR LF
pairs. Not \r\n or \n\n. Because those depend on your perl 
implementation and OS and EBCDIC usage. Lincoln's comments spell this
out. And I find it just as annoying, but how I feel about these
minor Perl / protocol / platform dependencies doesn't matter.

If we are writing to streams or files, particularly UNICODE content,
if we are on certain platforms, we need to remember to open the file
in binmode, as there is still some distinction between text/binary
modes on the filehandles in MS OSes.

The file separator code was a freebie in the CGI.pm code.
So, I think we are fighting the wrong fight maybe over the filepath
separator character stuff.

I'll agree to revisit this later when I get to the checking where
we spit out documents, get documents from URIs, and access
filesystem or stream objects.

Mike.
PS - I haven't worked on stuff this week cuz I'm swamped at work
and my WinCVS is busted. I think IE5.5 broke it. Something related
to MS having numerous undocumented versions of one 'MFC42.DLL' file.
Thank you Microsoft. :(