I'm in the same boat you are as far as trying to determine the best way to release this portalware I have.
I was thinking about encoder...
but if zend eats it's it's own dog food,
then I'd say the performance of the zend.com site is a classic example of what to expect with the use of encoder/optimizer.
Frankly, zend.com is my least favorite site to visit because the page has this unexplicable hang even after it's rendered on the page, sometimes up to 20 seconds.
I've been thinking about playing around with resources though and seeing if I could write a C program to decrypt mcrypt encrypted files and pass back a resource and then have PHP use the resource returned to do an include().
That would allow the files in question to be encoded without paying thousands of dollars for an app that will make your entire site run like a tortoise.
Essentially, it would be a software dongler
that would require a keyfile for access
and wouldn't really slow the PHP application down all that much if you were thoughtfull in your approach... and it would allow for non-incrypted files to be used as well.
Since my app is a portal framework, the people who use it would need use their own non-encrypted code to control it and they wouldn't need to get into my code since it's all abstracted.
But, I have been toying with the idea of just putting it under the GPL and posting the entire damn library on sourceforge or some other open source site for public participation. It definitely would be a useful piece of software for others to have access to since it allows you to host multiple sites of a single source distro and performs many mundane things like DB access, user login and Directory IO automagically.
And with a public project, you can often times wind up with a better codebase if you get some decent programmers to participate.
But on the other hand...
I really don't feel like managing a project like that since I have so many other apps to write.
So I'm still so very confused on this subject.