Best approach for temporarily swapping out a single perl module for a different version?

[nysus]

If I have a perl module installed on a development machine and I want to temporarily run a different version of that module (for example, a copy of the module in a local repo), I can modify the $PERL5LIB in my my bash profile and insert the path to the 'lib' directory in the temporary version of the module to the beginning of $PERL5LIB and then reload bash. Then, to restore the original module, I remove the path to the temporary module form $PERL5LIB and restart bash.

This isn't too bad but just curious to know if others do the same thing or if there is some less hassle free way of doing this. Perhaps I should be using local::lib instead? Note that I'm using perlbrew so I'm not sure if local::lib is a viable option for me. Thanks.

[choroba]

What I usually do in such a situation: I go to the ~/perl5 directory (created by local::lib), run

git init
git add .
git commit -m start

(if it's not already there, then I don't need to run anything).

Then I create a branch for experiments, e.g.

git checkout -b module-3.4

and actually upgrade the module.

Installing several different versions in different branches is possible, you can then use git to switch between them and compare the behaviour of your app or tests.

After the experiment is over, I either stay with the new version

git checkout master
git merge module-3.4

or I return to master and delete the branch.

I described the process when I first used it in my blog.

[nysus]

Let me see if I got this straight because I'm not familiar with local::lib and I'm using perlbrew which created ~/perl5 for me.

So basically you create a giant git repo for all the modules located in the root of ~/perl5? And then when you want to upgrade a single module, you create a new branch and modify that single module? Is that right?

[choroba]

Exactly. Adding a change can take a second or two, switching the branches is blazing fast.

[Fletch]

The git idea's pretty interesting/clever.

Then again nothing says you can't whip up a new local::lib sandbox directory and install your new stuff into that and prefix that new directory on PERL5LIB. That'll let you combine possibly upgraded dependencies and still use the normal tools to manage installing everything. If/when you decide to go with the new version you just go back and install into your "normal" l::l directory.

(And I really like the git idea.)

[jcb]

Other monks have mentioned a few solutions, but a big question is "how temporary?"

You could put the special version of the module in its own directory and use the -Idir option when you want to use that version. Then perl will look in that directory before the regular @INC, find the special version first, and load it.

[tobyink]

This is generally what I do.

[Flex]

I'm not exactly sure how your using the "insert the path to the 'lib' directory"
but use lib LIST; has a remove no lib LIST; or you can just restore the original @INC = @lib::ORIG_INC;. see lib.

[1nickt]

hi, try perlbrew lib.

Hope this helps!

[nysus]

Thanks for the tip but I couldn't figure out how to use it, exactly. So I create a new lib with perlbrew lib create blah and then I switch to it but then what do I do from there?

[nysus]

OK, think I see now. When you switch to new lib, it falls back to the lib for the installed version of perl in ~/perl5 if the module is not yet installed in the new lib. I think this might be a great solution. Glad I asked the question.

[Anonymous Monk]

This assumes "temporarily" means "only for a specific script." I put the "different version" in its own directory, then the script does a use lib 'different directory'; before it loads the module.