add automatic PyPI deployment #75

ltalirz · 2019-11-07T12:31:10Z

This automatically deploys any tags of a valid form v0.1.2 to PyPI using an encrypted PyPI API token (which provides access only to the optimade PyPI package).

I haven't tested this for this specific repo but I'm using the same approach in a number of other packages and I think it should work (one change: I used python 3.7 here to deploy, I guess this should be ok).

In case there is any issue, we can fix it when this is triggered with the next tag.

ml-evs

This looks useful, happy to accept pending my query

ml-evs · 2019-11-07T12:59:10Z

.travis.yml

+          secure: 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
+        on:
+          repo: Materials-Consortia/optimade-python-tools
+          all_branches: true


Do we want to push all branches or just master?

from my experience this can be useful for alpha/beta releases that one does not yet want to merge into master, but it comes from the gitflow model where you would have a develop branch for this.
up to you, I can also restrict it to master.

I just worry that everyone has push access to arbitrary branches except master at the moment, (maybe even to master, need to check), so someone could hijack the PyPI version with unreviewed code. (I'd also be fine with moving to the gitflow model!)

CasperWA · 2019-11-08T13:50:23Z

Great! Thanks @ltalirz.

This automatically deploys any tags of a valid form v0.1.2 to PyPI (...)

I was wondering if it would be possible to have the form v0.1.2.3? Then we could have the first three be major.minor.patch of the OPTiMaDe spec., and the last be the specific revision for this repo.

ltalirz · 2019-11-08T13:58:25Z

I was wondering if it would be possible to have the form v0.1.2.3? Then we could have the first three be major.minor.patch of the OPTiMaDe spec., and the last be the specific revision for this repo.

It's certainly possible but I don't think it is a good idea to couple the version numbers of optimade with the version numbers of this repository.
I would vote to stick to semantic versioning for this package in order to be able to signal bug fixes / new features / breaking changes.

ltalirz · 2019-11-08T14:02:36Z

I certainly agree, though, that it is important to clearly display, which optimade version the python tools are implementing.
It should be possible to get this information programmatically from python, and perhaps we should even display it in the README of the repo?

CasperWA · 2019-11-08T14:07:39Z

It's certainly possible but I don't think it is a good idea to couple the version numbers of optimade with the version numbers of this repository.

This is exactly what was proposed in #67, but to have only major.minor be equal to (not patch).

I would vote to stick to semantic versioning for this package in order to be able to signal bug fixes / new features / breaking changes.

I don't know. I would like to have both 😅 Whether this be via v(OPTiMaDe spec version).(optimade-python-tools version) or something of those lines? Would that be too much?

I see this repository mainly for the value of the pydantic models, which may be used by any implementation to validate your response.
The server part of this I see merely as an example of how such an implementation can be done.
Maybe it would make more sense then to split the repo up, keeping the models (and soon to come external API validator) in this repo and then a "test" server implementation in another repo?

ml-evs · 2019-11-08T15:49:57Z

We should also consider how we're going to support multiple versions of the spec at once, e.g. validate implementation against v1.0 or future v2.0. This is worth a separate issue but would definitely effect the version numbers for this repo.

Do we a) manually version the models inside this repo (e.g. models/latest, models/v1.0, models/v1.1, b) split the models into a separate package so switching version is just a case of pip install optimade-python-models==0.9.7 then have that extra package exactly follow the spec version, or c) something better?

I guess the example server code will only ever track the "latest version" of the spec (or should it go in a separate versioned package with the models?), but the validator/grammar should be usable against arbitrary versions...

fekad · 2019-11-08T15:51:06Z

It's certainly possible but I don't think it is a good idea to couple the version numbers of optimade with the version numbers of this repository.

This is exactly what was proposed in #67, but to have only major.minor be equal to (not patch).

I would vote to stick to semantic versioning for this package in order to be able to signal bug fixes / new features / breaking changes.

I don't know. I would like to have both 😅 Whether this be via v(OPTiMaDe spec version).(optimade-python-tools version) or something of those lines? Would that be too much?

I see this repository mainly for the value of the pydantic models, which may be used by any implementation to validate your response.
The server part of this I see merely as an example of how such an implementation can be done.
Maybe it would make more sense then to split the repo up, keeping the models (and soon to come external API validator) in this repo and then a "test" server implementation in another repo?

Please correct me if I'm wrong but the GitHub tag is completely independent from the version number used by pypi and defined in setup.py. Of course, we can use the same version for both but it has to be modified at the same time...

ltalirz · 2019-11-10T20:15:49Z

To me, the point by @ml-evs above concerning support for multiple API versions is just another reason not to couple the versioning of the package to an API version. From my point of view this PR is ready to be merged.

Please correct me if I'm wrong but the GitHub tag is completely independent from the version number used by pypi and defined in setup.py.

Correct.

Of course, we can use the same version for both but it has to be modified at the same time...

Not quite sure what you are trying to say here...
anyhow, from my point of view there is I see no good reason to use different conventions for tags (which will be shown as "releases" on github) and releases on pypi.

ml-evs · 2019-11-11T14:15:30Z

Cheers @ltalirz , I guess we can move the versioning discussion to an issue

ml-evs · 2019-11-11T14:24:28Z

Travis isn't running on this branch for some reason...

ltalirz · 2019-11-11T15:16:36Z

Feel free to merge whenever you see fit.
Note that the extra travis job won't show up here (only in the build triggered by pushing a tag of the right format).

ml-evs · 2019-11-11T17:06:23Z

Feel free to merge whenever you see fit.
Note that the extra travis job won't show up here (only in the build triggered by pushing a tag of the right format).

The actual PR isn't showing up on travis so it can't be merged (unless we change our settings). Do you have any strange permissions on your fork?

ml-evs · 2019-11-11T17:11:43Z

Temporarily changed settings and merged as there was no way to trigger a travis build.

ml-evs · 2019-11-11T18:59:46Z

It seems that this PR actually breaks normal travis builds. (Unfortunately travis status was reporting some issues when I was merging this, so just assumed it was that...)

I can't even trigger a build with a custom config that excludes the changes in this PR, so really not sure what is going on.

fekad · 2019-11-11T19:17:21Z

Thanks, I will have a look at it. Yesterday it was completely fine ( https://travis-ci.org/Materials-Consortia/optimade-python-tools/builds/609928631 )

…

On Mon, 11 Nov 2019 at 18:59, Matthew Evans ***@***.***> wrote: It seems that this PR actually breaks normal travis builds. (Unfortunately travis status was reporting some issues when I was merging this, so just assumed it was that...) I can't even trigger a build with a custom config that excludes the changes in this PR, so really not sure what is going on. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#75?email_source=notifications&email_token=ABTPS7NWVMJZ62CPJRLQTWDQTGTSHA5CNFSM4JKF5H62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEDXYX6I#issuecomment-552569849>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTPS7M2JYIRE622TJHA7KTQTGTSHANCNFSM4JKF5H6Q> .

-- *Ádám **Fekete | *adam@fekete.co.uk

ltalirz · 2019-11-11T19:24:26Z

@fekad I think that's on me here - very sorry about this.
I'll look into what I did wrong here...

ltalirz · 2019-11-11T20:07:11Z

It seems I ran into this slightly weird travis issue: travis-ci/travis-ci#8536

My other packages use builds on multiple python versions/with different environment variables, and there you can keep your tests defined a the top level, with the deploy stage as a job.
When you have only one build (as we have), for some reason you need to specify the tests as a separate job (see build on my fork: https://travis-ci.org/ltalirz/optimade-python-tools/builds/610523570).

So, we have two choices:

I move the tests to a job, as done here: https://github.com/ltalirz/optimade-python-tools/blob/add-pypi-deployment/.travis.yml
we start testing on multiple python versions , e.g. python 3.8 as well ;-)

@ml-evs Let me know what you prefer

ml-evs · 2019-11-11T20:39:33Z

Looks like fun... I guess we may as well test python 3.8 too, but I'm not fussed either way!

add automatic PyPI deployment

a37f46e

ltalirz requested a review from ml-evs November 7, 2019 12:31

ml-evs requested changes Nov 7, 2019

View reviewed changes

ltalirz added 2 commits November 8, 2019 15:03

restrict deployments to master branch

c72a55e

Merge branch 'master' into add-pypi-deployment

73618fc

Merge branch 'master' into add-pypi-deployment

44f225d

ml-evs approved these changes Nov 11, 2019

View reviewed changes

ml-evs merged commit 187c1f8 into Materials-Consortia:master Nov 11, 2019

ltalirz mentioned this pull request Nov 11, 2019

fix travis build #77

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add automatic PyPI deployment #75

add automatic PyPI deployment #75

ltalirz commented Nov 7, 2019 •

edited

ml-evs left a comment

ml-evs Nov 7, 2019

ltalirz Nov 7, 2019

ml-evs Nov 7, 2019

CasperWA commented Nov 8, 2019

ltalirz commented Nov 8, 2019 •

edited

ltalirz commented Nov 8, 2019 •

edited

CasperWA commented Nov 8, 2019

ml-evs commented Nov 8, 2019

fekad commented Nov 8, 2019

ltalirz commented Nov 10, 2019

ml-evs commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ltalirz commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ml-evs commented Nov 11, 2019

fekad commented Nov 11, 2019 via email

ltalirz commented Nov 11, 2019

ltalirz commented Nov 11, 2019

ml-evs commented Nov 11, 2019

add automatic PyPI deployment #75

add automatic PyPI deployment #75

Conversation

ltalirz commented Nov 7, 2019 • edited

ml-evs left a comment

Choose a reason for hiding this comment

ml-evs Nov 7, 2019

Choose a reason for hiding this comment

ltalirz Nov 7, 2019

Choose a reason for hiding this comment

ml-evs Nov 7, 2019

Choose a reason for hiding this comment

CasperWA commented Nov 8, 2019

ltalirz commented Nov 8, 2019 • edited

ltalirz commented Nov 8, 2019 • edited

CasperWA commented Nov 8, 2019

ml-evs commented Nov 8, 2019

fekad commented Nov 8, 2019

ltalirz commented Nov 10, 2019

ml-evs commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ltalirz commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ml-evs commented Nov 11, 2019

fekad commented Nov 11, 2019 via email

ltalirz commented Nov 11, 2019

ltalirz commented Nov 11, 2019

ml-evs commented Nov 11, 2019

ltalirz commented Nov 7, 2019 •

edited

ltalirz commented Nov 8, 2019 •

edited

ltalirz commented Nov 8, 2019 •

edited