I've been brainstorming for a while regarding the best approach to distribute closed-source modules. I'd like to mention that I have found similar question(s) on SE sites but don't think this is exactly the same scenario.

Let's say I wrote a set of well-tested PHP classes which I used as a core for multiple projects (copyright is mine). It has EntityManager, ACL, ORM, Configurable Cacheing, Action logging Admin panel functionalities etc which can be very handy in any database driven projects. (Let's not start argument regarding the integrity of the project itself ;)

My plan is to implement SAAS with the current set of functionalities, but the problem is most of the people I discussed would like it hosted in their servers and wouldn't mind paying a small recurrent loyalty fee. That leads to the ultimate question of choosing the best way of distributing such module / micro-framework.

I found following options to consider:

PHP Extension: First solution that'll come to anyone's head is to develop and distribute as a standard php extension (Which will check for valid license-code in the cache once in a while and a cronjob will update license from license server periodically). I checked php-cpp.com regarding this and it seemed pretty impressing. One major difficulty is that the extension should be able to talk to mysql / redis etc (based on the configs passed to instantiate the object) by itself. At the moment, current php classes utilise MySQL and PHPRedis extension. I'd like to be able to add Mongodb, Cassandra/ElasticSearch functionality in near future. I'm not sure how it could be achieved. If anyone can possibly shed a light about how an extension can perform mySQL query internally that would be a great help.

Zephir: I've seen some performance benchmark and Phalcon is one of the top competitors. The major advantage for me is that Zephir is php-like and it'll take considerably less time than writing C++ extension. But looking at the discussion boards, it appears that Zephir is a bit undercooked. By the time it'll be ready for production, php itself will introduce PHP7/NG. Also, same difficulty in the extension writing is present here.

HHVM: Arguably, the best solution as it converts everything to byte-code and I wont't have to worry about the extension-difficulty here. But I have a new problem: I won't be able to allow users to attach their function (anonymous function) to the events triggered in my module (e.g: OnDuplicate() or OnPost()).

Any idea, resource or comment is greatly appreciated. Thanks.

  • What are you trying to achieve? Why can't you just use plain PHP code?
    – Craig
    Jan 19, 2015 at 21:24
  • A highly configurable extension-like module to distribute to my potential clients (mostly other media agencies/developers). I may be a bit paranoid regarding source-code distribution but still I believe close-sourced solution has certain advantages in this scenario.
    – user123808
    Jan 20, 2015 at 9:14

2 Answers 2


I honestly think you need to give up on not distributing code in your case. Closed source does not mean you don't/can't distribute code, it means that they are not allowed to distribute it.

I think you should really evaluate your goals. If you want to stop people from looking at your code, then use an obfuscateor, if you want to bill per install then use a licence checker of some kind. If you can't trust your clients, then get new clients.

Besides (and I mean this in a nice way), your code is "worthless" in this type of scenario, it's your idea and implementation that has value. Even if someone never gets your source code, they can just copy the idea.

  • +1 for stating the obvious difference between 'idea' and 'worthless code'. I am a sucker for simplicity and distributing codes seem to be the simplest and a good way to move forward. However, I'm still not sure if it's the right way because of 3 core reasons: a) Couldn't find a similar (Saas) service that distributes code and supports 'em. b) Once the code is available in the internet, I have near-zero intension to spend my time chasing the culprit. Instead I'd much rather prefer to bulletproof the license agreement and spend my time improving it. c) Compiling to byte-codes makes it faster.
    – user123808
    Jan 23, 2015 at 17:15
  • c. I can't argue with. That's one good reason to compile (language dependent). id soft (quake) distributes their source code and is closed source. Can't comment on SaaS that distributes that is similar to yours but Shopify is open source and they make good money.
    – coteyr
    Jan 24, 2015 at 3:07
  • I suppose you could do a compiled core, and then a code based interface. The "others" could call the code based interface, which they could see and overload etc. etc. But the core could be compiled. Nvidia does something like this with their Linux driver.
    – coteyr
    Jan 24, 2015 at 3:10
  • Not sure if I understood the connection between compiled core and coded base interface. Let's say I compiled the application using hhvm with fast-cgi; I'd only be able to use it by url. It'd be great if I could possibly call class/method individually. Did I miss anything?
    – user123808
    Jan 25, 2015 at 19:13

What about using Zend Guard or ionCube PHP Encoder? Or why is it a problem to distribute your source code for a fee?

This is a problem not special to PHP. Many companies write libraries that they sell in source code form.

  • Thanks for your response. I've checked ZendGuard and ionCube. IMO, using services like these add extra layer of complexity to development environment. I'd prefer to have more work done (upfront) from my side than spending time on support. Also, I am open to suggestions if there's alternative solution in Python.
    – user123808
    Jan 18, 2015 at 0:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.