On 2 Nov 2004, at 23:22, Paul McCann wrote:
Hi Richard, you asked...
I'm about to write a number of fairly simple perl scripts. They'll be reading xml files. Unfortunately they've got to be installed on a client machine that uses 10.1.5, owned by a fairly untechnical user. I've always had control of the environment before, and I've no experience with making installation nice and easy. I'm fairly sure developer tools won't be installed on the client machine.
I don't need much of a gui interface, so I'm planning on a drop-script front end. What's worrying me is getting the modules and parser installed.
How can I find out what is installed on the client machine by way of xml parsers? Did 10.1.5 come with expat or libxml? If so, how can I make installing XML::LibXML or XML::Twig easy? I might be able to make a 10.1.5 environment temporarily to test this out - will this make my life a lot easier?
Ouch: I don't think either expat or libxml2 will be present on the raw
system. And compiling libxml2 for 10.1.x was, let's say, "interesting".
You can get old binaries (via dmg's) of libxml2 and libxslt for 10.1.2
that should work fine in 10.1.5 from
http://www.xmlperl.com/modules.php? op=modload&name=Downloads&file=index&req=viewdownload&cid=1
Note however that these are pretty old versions: 2.4.15 and 1.0.12.
But that still leaves the difficulty of getting XML::LibXML or some of
the expat dependent modules to install. And that's almost certainly
going to require a 10.1.x machine with dev tools, and maybe an older
version of the modules. (Probably depends which route you take: libxml2
or libxslt.) I think you will want to make a 10.1.5 machine to do this.
Doubtless you've thought along the lines: "wouldn't it just be easier to
upgrade the client to 10.2 or 10.3?"
It would indeed. I've also considered using regexen instead of an xml parser - its simple enough that I could get definitely get away with it.
Anyway, I'm rambling: if you set up your 10.1.5 machine it's probably worth trying to compile libxml2 and/or expat directly from source so as to get recent versions. Even then you'll probably end up having to dump your constructed perl directory onto the client machine, which is a little gross. Otherwise you might try the above binaries.
I think it will depend if I can create such a machine in the time frames needed.
Best of luck!
Looks like I may need it.
Thanks, richard
Paul