cturbelin/rserve-php

Session test in fork doesnt pass

shadiakiki1986 opened this issue · 4 comments

shadiakiki1986 commented

Hi. I made another fork of your repository with a major refactor to use php namespaces and composer's autoloading. I also added a travis file for automatic Continuous Integration. The only problem I have is that the Session test doesn't pass. Can you take a look at it in my fork so I can fix it and make a pull request?

Here is my fork, and here is the travis log showing that all tests pass except for the session test which I marked as skipped

cturbelin commented

Hi !

Thank you but this work has already been done in the 2.0 branch of the repo
(also with composer support) published several months ago. It will be
released soon, when the tests are finalized.

I will look your code to see if I can integrate back some changes you
proposed, but I could not merge this branch with yours since many changes
and fixes have already been done on this new branch (refactoring, choice of
namespace, and so on).
Please use this branch instead of yours, because all tests and next releases
will be based on this branch.

Best,

Clément

shadiakiki1986 commented

Ah ok, my bad. I was looking at the master branch. Thanks!

shadiakiki1986 commented

Ok, so I branched off 2.0 and re-integrated my edits, except for the ones having to do with refactoring for psr4 autoloading since it's already done in your branch.

I ended up again at the session test not working. Here are my links again:

  • fork of 2.0
  • travis logs for 2.0 showing Connection refused. I noticed that the session's port number is different than the original port number (6311 as in the config.php.sample)

Any ideas about this before I make a Pull Request?

cturbelin commented

Thank you very much for all the work. I've just merge the pull request (after a long time, I know, but the project is not abandonned !)
I'm still stuck with the session problem.
Using Rserve 1.7.3, php 5.6.36
Yes Rserve listen on another port for the session (I've seen this will be removed on 1.8 version).
The result of the test are not predictable : sometimes it passes, sometimes not. It seems that Rserve responds the packed is malformed and close the connection (so the socket is closed). But I dont understand why it is not reproductible yet.