[Python-Dev] API for bugs.python.org or some statistic about lib/module merge rate?

Louie Lu me at louie.lu
Fri Apr 28 00:12:27 EDT 2017


2017-04-28 4:06 GMT+08:00 Terry Reedy <tjreedy at udel.edu>:
> On 4/27/2017 7:57 AM, Louie Lu wrote:
>
>>> Lib/idlelib/idle_test/test_calltips.py has 1 and 1
>>> bpo issues:     20185 22392 24665
>>> None of these are idlelib issues.
>>>
>>
>> Not that directly, but these bpo issue PR has changed the file.
>> See: bpo-20185: PR 542 bpo-24665: PR 89, and a issue fixed but close PR
>> 898.
>
>
> #20185: The PR number is *very* helpful.  This issue has multiple patches
> and 5 PRs merged to default.
>
> #22392: PR 89 has about 10 commits.  The original, easy-to-find patch,
> https://github.com/python/cpython/pull/89/commits/6ea78e33b599605b0c1c35fbbc2df493d6dfc80a
> did not touch test_calltips.
>
> It is touched by the latest, not so easy-to-find commit
> https://github.com/python/cpython/pull/89/commits/54de7aa6c6fffe9b2248153051a24b0e658bf665
>
> I don't know how to get the complete merged diff that would be applied if
> the green button were pressed.
>
> #22392 has PR 899 (merged) and PR 898 (closed).  The latter was a blunder
> that incorporated about 100 already applied patches, mostly from 2016.  Take
> a look.
> https://github.com/python/cpython/pull/898
>
> Part of doing good statistics is removing bogus data from datasets. Even
> though you cannot remove 898 from github, you should ignore it in your
> analysis.  All the hits it generates are false positives.  This is probably
> the source of many other false hits for idlelib.
>

Now add the blacklist for PR, this is still manual to add into the config
I have three cases: PR 898, PR 703, and PR 552.

The "automatic" identified should be add in future (probably)

>> The detail now will show the relative PR (which bpo and which PR that
>> is merged, fixed, or open) inside.
>>
>> e.g.
>> Lib/idlelib/idle_test/test_calltips.py
>> bpo issues: 20185 22392 24665
>> PR:             bpo-20185: PR 542 bpo-24665: PR 89
>
>
> Note that 898 is so bogus that it cannot be handled properly.
>
> For me, the immediate use case is to find patches to idlelib by others that
> I am not aware of, like the two real ones to test_calltips.  I would like an
> entry box for the start date to put in date last checked. What would be

The latest bpo-issue check date will now be shown when the cursor
hovers on the issue.
(That mean, mobile can't see the date now. I think (on mobile) this
isn't that needed)

> really super would to be able to exclude issues (and associated patches)
> assigned to me.

This may be update in the future. (Assignee from bpo or GitHub?)
>
> When you get this polished, I hope you open an issue and PR to add it to one
> of the cpython and mention it in the devguide.
>

Yes, I will. A small question what does it mean "add it to one of the cpython"?

>
> --
> Terry Jan Reedy
>

Best Regards,
Louie.


More information about the Python-Dev mailing list