Plans for release of pika from current develop

[Martin Uhrin]

Dear pika developers,

I wanted to check what the release schedule is like for pika 1.0 (or at
least a release from the current state of develop). I was using pika-dev
on pypi for a while but it has quite a few issues that have now been fixed.

If there isn't a plan to release 1.0 any time soon, would it be possible
to at least release a new beta as pika-dev on pypi? This would mean I
could pip install my libraries easily (with dependency links).

Background info:  I use pika through a comms library I wrote (topika -

https://github.com/muhrin/topika/ - a backport of aio_pika to support
python 2.7), but I rely on recent functionality in pika on the dev branch.

Thanks,

-Martin

[lba...@pivotal.io]

Hi Martin,

There is no schedule for Pika 1.0, but I will make time to release Pika 1.0.0b1 this week.

Thanks

Luke

[Martin Uhrin]

Thanks Luke, I'll keep an eye out.

-Martin

> --
> You received this message because you are subscribed to a topic in the Google Groups "Pika" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/pika-python/UQO0h5aBXzU/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to pika-python...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[lba...@pivotal.io]

Hi Martin,

Please give this release a try:

https://pypi.org/project/pika/1.0.0b1/

Thanks,

Luke

[Rich Wood]

We have been using pika in a production system for approximately 6 months - we are running on ephemeral EMR clusters and when a new cluster starts it gets the latest version using easy_install - today we started getting the 1.0.0b1 release and basic connection functionality stopped working - for now we're going to hard code getting version 0.11.2 but needless to say we're less than happy with this result.

Here is the message we got with the new version:

Connecting to amqp://xxxxxxxxxxxxxx@xxxxxxxxxxxx:5672/%2F?heartbeat_interval=30

Exception in thread Thread-1:

Traceback (most recent call last):

  File "/usr/lib64/python2.7/threading.py", line 804, in __bootstrap_inner

    self.run()

  File "/usr/lib64/python2.7/threading.py", line 757, in run

    self.__target(*self.__args, **self.__kwargs)

  File "dlp_builds_processor.py", line 194, in start_process_work_received_monitor

    rmq_con.run()

  File "/home/hadoop/dlp-spark/utils/rmq_comm.py", line 387, in run

    self.open_connection()

  File "/home/hadoop/dlp-spark/utils/rmq_comm.py", line 378, in open_connection

    self._connection = self.connect()

  File "/home/hadoop/dlp-spark/utils/rmq_comm.py", line 91, in connect

    stop_ioloop_on_close=False)

TypeError: __init__() got an unexpected keyword argument 'stop_ioloop_on_close'

[Martin Uhrin]

Dear Rich,

sorry to hear you're having problems - I feel partly responsible given
that I asked for the new pre-release.

So, I don't know easy_install very well but what's done is just take
the list from https://pypi.org/simple/pika/ and downloaded the latest
version (1.0.0b1). Pika 1.0 will make breaking API changes (w.r.t.
0.x versions) without which certain operations are almost impossible
(which is why I use it in `topika`).

pip will do the right thing: If you do `pip install pika` it will not
install a pre-release version (i.e. version numbers with a/b/rc, etc
in the string) and just give you 0.12 (that said when 1.0 proper is
release you would have the same issue). I don't know if it's possible
to do this with easy_install. If not I would recomment keeping your
current solution or migrating to a pip style setup.py.

-Martin

[lba...@pivotal.io]

Hi Rich,

Pika 1.0.0 will contain breaking API changes as allowed by semantic versioning. These changes will be documented and announced. I haven't announced 1.0.0b1 on this mailing list, rabbitmq-users nor the GitHub release page as this is a beta release and I am still putting together the announcement.

Martin's suggestion to use pip is a good. I don't know why your system didn't auto-update to Pika 0.12.0 as that is the latest stable release.

In any case, automatically updating to beta releases should be avoided, especially if the major version changes.

Thanks,

Luke

[Rich Wood]

Thanks guys for your replies. We're using easy_install because of issues running pip as sudo on EMR nodes. We solved the problem with easy_install for now by hardcoding the version to use but I guess we'll have to take another look at getting pip to work after modifying our code to work with the version 1 api.

rich