Use checkout on 5.2, normalise http-examples [7xx] #1218
Conversation
This was already done correctly for the other Plone versions. Done for all versions in the same way now: first extend the dist.plone.org versions and then extend base.cfg, which unpins plone.restapi. Also: test with latest Plone 5.2.5. And remove pins that are already in 5.2.5 with a good version.
When writing the http-examples, use the original simpler generated schema name.
The first part of save_request_and_response_for_docs was an exact copy save_request_for_docs (before my change in the previous commit).
…mparison. This avoids many uninteresting differences.
This avoids differences when running from the Plone coredev buildout, which uses a random port. A few differences remain, at least when running the test in Plone 6. On Plone 5.2, only workingcopy_baseline_get.resp has a difference, and this depends on your timezone.
These are only whitespace changes.
|
@mauritsvanrees thanks for creating this Pull Request and help improve Plone! To ensure that these changes do not break other parts of Plone, the Plone test suite matrix needs to pass. Whenever you feel that the pull request is ready to be tested, either start all jenkins jobs pull requests by yourself, or simply add a comment in this pull request stating: With this simple comment all the jobs will be started automatically. Happy hacking! |
|
@jenkins-plone-org please run jobs |
This is no longer needed, since we update normalize the http-examples output in test_documentation.py. Also, without hardcoded port, it becomes possible to introduce tox and run it in parallel, if we want.
…expected name. Uglier, but closer to reality.
code-analysis did not complain, so I assumed it was fine, but apparently not.
| document_schema_re = re.compile( | ||
| r"^plone.dexterity.schema.generated.plone_.*_Document$" | ||
| ) |
There was a problem hiding this comment.
I think we can be less invasive here. That is, replacing only the really dynamic part. I suggest:
| document_schema_re = re.compile( | |
| r"^plone.dexterity.schema.generated.plone_.*_Document$" | |
| ) | |
| document_schema_re = re.compile( | |
| r"^plone.dexterity.schema.generated.plone_5_\d*_2_\d*_0_Document$" | |
| ) |
So, if we have changes like what happened in plone.dexterity 2.10+, the tests will fail.
There was a problem hiding this comment.
Why would we do this?
I have no idea if the 5, 2 and 0 are always here or if they can differ as well. Seems pretty random to me.
This is an internal dexterity detail. If it changes, I do not want these tests to fail.
There was a problem hiding this comment.
They will only change if we have structural changes like the ones that occurred on plone.dexterity 2.10. They don't change "via parameter". In this case, I find it valid for the test to fail, as failed. If we had this regular expression, before the change from plone.dexterity 2.10, the tests would not fail, even with the name pattern changing. And I think that in this case it should fail.
schemaName after this line:
it's always something like:
plone_5_1631640310_2_6213937_0_Document
There was a problem hiding this comment.
I see the 5 and 0 come from the SchemaNameEncoder.
I see no value in having a test in plone.restapi fail when this class changes.
There was a problem hiding this comment.
We can have more similar values to the real ones in the documentation. I think that's why we have testing to generate documentation, rather than just writing the documentation.
There was a problem hiding this comment.
Fine, since no one stepped up to argue in my favour, I have changed it: commit 8921bec.
I think it is fine to merge now, right? Well, once everything is green.
|
@jenkins-plone-org please run jobs |
Will do. |
mauritsvanrees
deleted the
maurits/use-plone-restapi-checkout-52-7xx
branch
|
Done: #1226 |
This is a backport of my PR #1215 to branch 7.x.x.
It also updates 5.2.x to use Plone 5.2.5, including using the newer
plone.dexterity. We normalise its unpredictable behavior schema names in the http examples.This PR is best viewed with
?w=1at the end of the url, to ignore whitespace changes, especially with the commit that updates thehttp-examples, which is actually a whitespace-only change.