Pum integration2

[sylvainbeo]

@m-kuhn @haubourg Hi, have a look to this PR. I've added pre and post-all delta file + a pre for 1.2.8.

Please comment.

[3nids]

is that a green light ???? haven't seen one for quite a while around here... good job to all of you!!!

[m-kuhn]

Great stuff, that's good to see.

@marioba any feedback from your side concerning the use of os.system? Alternatively there would be the subprocess module which could also intercept stdout and stderr, should that be sent back to Pum, and of course psycopg. Not sure how you envisioned this.

[marioba]

👍for the green light...

In the DeltaPy abstract class, there is the write_message method designed to write the subclass outputs. Currently, the method doesn't make much more than a simple print() but I think it would be good to have a centralized output so, if we want to manage it differently in the future, it's easy to modify it. And maybe it would be also a good idea to define some exception types that the subclasses can raise in the run method...

[m-kuhn]

Thinking about this again, it might be nice to send all sql through psycopg2. At least from a transaction perspective it would be nice to send everything through the same interface and have total control to commit or rollback.

[sylvainbeo]

Ok all, we wait for @haubourg to have a look, and we merge.

[ponceta]

Thanks for the good work, this will be great to have this!

There's a small error here :

https://travis-ci.org/qwat/qwat-data-model/builds/293131379#L2870

Probably linked to SRID not beeing initialized somehow?

Il like the green tick but having errors and still been green can be a false positive for me.

The auth token is not working too?

https://travis-ci.org/qwat/qwat-data-model/builds/293131268#L5398

[sylvainbeo]

@ponceta Ah oups. did not see this one, because travis was collapsing the lines. Yes indeed, it is because of the missing SRID. I have a look.

[m-kuhn]

Can someone check if the GH_TOKEN has the public_repo permission set?

[m-kuhn]

@sylvainbeo I guess you need to check the return value of the psql call.

[sylvainbeo]

@m-kuhn The problem is on the post-all process. No sql is launched directly, just .sh I have a look.

[sylvainbeo]

SRID problem seems to be solved.

Now i see another one 'schema "qwat_od" does not exist'

[sylvainbeo]

• @m-kuhn

Traceback (most recent call last): File "./.deploy/create_release.py", line 154, in main() File "./.deploy/create_release.py", line 144, in main url='{release_url}?name={filename}'.format(release_url=release['upload_url'][:-13], filename=filename) KeyError: 'upload_url'

[m-kuhn]

@sylvainbeo , that normally happens when there's something broken with the api token

Can someone check if the GH_TOKEN has the public_repo permission set?

[m-kuhn]

That should give better output: https://github.com/qwat/qwat-data-model/pull/206

[sylvainbeo]

I'm not a GH_TOKEN expert. here is the configuration for the tken (qwat_admin user)

[m-kuhn]

Looks good I think

[m-kuhn]

Ah wait... Actually deploy should only happen on tags... We didn't have a tag here, right?

[sylvainbeo]

@m-kuhn Indeed no tag.

[sylvainbeo]

Anyone has an idea on " schema "qwat_od" does not exist"..very weird.

[sylvainbeo]

mmmm... i have one. Maybe the .sh scripts are not launched on the right DATABASE.

[sylvainbeo]

Fixed !

[sylvainbeo]

Question: in the post-all.py file, i specify qwat_test as the output DB... maybe this should be a param.

[haubourg]

Merging that at last!

[m-kuhn]

:tada: