Arch Linux User Repository

Am I the only person thinking a package submitter (cough cough) saying 'I have an update, give me ownership to receive it' to its current, active maintainer instead of simply posting/sending it is kinda hostile?

Fails to build:

test_enforce_content_length_no_body passed 1 out of the required 1 times. Success!
test_pool_size_retry_drain_fail passed 1 out of the required 1 times. Success!
test_ignore_broken_pipe_errors passed 1 out of the required 1 times. Success!
test_multipart_assert_header_parsing_no_defects passed 1 out of the required 1 times. Success!

===End Flaky Test Report===
==== 3 failed, 1628 passed, 286 skipped, 56 deselected, 148 warnings in 265.04 seconds ====
==> ERROR: A failure occurred in check().
    Aborting...
error: failed to build 'python2-urllib3-1.26.15-1': 
error: packages failed to build: python2-urllib3-1.26.15-1

Maybe to avoid situations like the one of yesterday morning which caused the deletion of ipython2 instead of its fix, I should inform 'meticolous' user developers whenever I have any new update on something regarding this.

Really, I will keep you informed at every step so you won't have to file a new complaint every time you may think I am idle.

I need to stress the above is an ironic hyperbole; yet, I still think you need to know about my workload at this particular time.

Monday 29th January 2024

08:15 am

I have forked python-tornado repo.

8:36 am

I have finished parameterizing the upstream pkgbuild so one has to only once write the package name to potentially switch module and python version.

As usual since last year I have also inserted new contributors names, licensed the package and reduced the length to less than around 20 columns, so everything is readable and easily editable from a mobile portrait device.

Also I have inserted 'new' supported architectures so people can expect to know if a certain package has been tested on a certain platform and worked at the time of publishing given available installed libraries.

8:50 am

I have double checked the original upstream package I've forked and pushed and I will now build it.

Later, I will do another clone and see the diffs with the last time the python2 version of the package has been built using repo log.

8:58 am

The device I use to write here has discharged, it will have to stay under charge for a while before resuming. Meanwhile I do start the build.

9:02 am

The 5 minutes long build for python-tornado on the device I am using ended successfully so I am proceding to simply switch python to python2 in the _py variable and see if it builds.

9:05 am

While cloning python2-tornado, I remember the original repository is still there, because actually aur deletions just disable the web interface for the package, while keeping it available (but not discoverable) to everybody.

I also do remember the same person complaining for the missing dependencies of this package may be the same person who deleted python2-tornado, as I have already seen that happening more than once, for example with some ipython2 dependencies which had been deleted to justify deleting their parent.

So to avoid that weird character, this time I take the special care of fixing all the packages in a single step so he won't see any red on any page and maybe I will spend a nice day of unpaid contributions with my anonymous maintainer non-friends who love to press labeled buttons instead of writing, talking or replying to emails.

9:42 am

I have finished upgrading python2-tornado same way I did with its Python 3 counterpart and I am uploading it on its (temporary) ur http mirror.

I start building it on my device.

10:36 am

I just had dinner. I'll try to go to sleep now maybe.

2:44 pm

I wake up; I drink some coffee and read new messages on the telegram arch group I help with.

I will chill out another couple minutes before resuming the never ending unpaid task of being useful to yourself and to others.

3:00 pm

I resume building python2-tornado. I remember there are no others tornado packages so I also add a providefor that and tornado2.

3:14 pm

Build is working, I've pushed last working commit and now it's time to obtain and check runtime dependencies, starting with python2-futures.

3:26 pm

python2-futures builds. I add a couple stylistic changes, license and mirror it on the temporary ur namespace. I will then have to comment on aurweb about the new mirror but without reporting the package out-of-date, so simply notifying its maintainer somebody made an update available whenever he feels like it's appropriate to update it, given it's almost archive and if he built the package at an earlier time it will still keep working.

In fact, most people who still use python2 still have all the original recipes and binaries on their machines from [main] and [extra], including myself, if I had my main machine available.

3:53 pm

I have built and updated a python2-tornado dependency, python2-singledispatch.

I discover python2-futures maintainer is me, so I thank myself as usual and push the upgrade.

4:04 pm

I discover python2-singledispatch, dependency of python2-tornado, has been deleted by same person who has also deleted python2-tornado and who is currently complaining this package misses dependencies, so really exhibiting a very peculiar behaviour.

4:23 pm

I upgrade, fix and be ready to republish python2-backports-abc, dependency of python2-tornado.

5:00 pm

I have updated the style of python2-urllib3 and verified it's working so I am starting to push completed work.

I receive the first out-of-date notification for a package I have just finished updating, python2-futures.

So I do really think something has to be done about Marcell's bias towards me if he is so obsessed to mark out-of-date a package literally seconds after pushing it and complaining about the pushes I am gonna do from there to 10 minutes.

Essentially because users can't go check how with his notifications he manages to be very very aggressive in his non-communications.

Like when I adopt a package which has been abandoned for years and after a couple minutes I have him writing to me as if I had been its maintainer all the time since then.

Oh. And I understand the above, given users can't know when others have adopted packages. That's also why it offends me he said what he said yesterday.

Clearly I can't know how he behaves with others, this is what happens to me though. What strikes me though is how he can't simply say what packages he consider higher priority instead of letting me guess.

I also don't like seemingly strategic reports to seemingly make me lose time while I am doing other things on here, which has seemed to me happening a couple times.

In general I try to always assume good faith and it seems pretty clear to me. Feedback appreciated.

5:34 pm

It seems the reporter from above longs for tests in python2-futures which were not present not even when the package was in [extra].

I wrote a basic test invocation and updated it, since he can't even wait for me to finish this. I do wonder if his friends dedicate him as much time as I dedicate to him.

5:42 pm

I'd like to highlight I've tried to break ice with Marcell over time by sending him some instant messages with technical opinions and roadmaps some time ago but he never replied.

On the opposite, I can't possibly understand how it is we are not best frens considering how much time we both spend on here.

6:09 pm

I have finished building python2-trustme but before pushing I first want to check its dependencies are working.

6:47 pm

I finished building and upgrading python2-idna, needed by python2-trustme.

I will have to build a ton of arm packages tonight so I may be about to fix python2-bcrypt.

8:22 pm

I have published python2-bcript. While the mobile router which recharges slower than it discharges, I will now go buy some groceries and then eat.

Tuesday 30th January 2024

0:30 am

I resume working from python2-ply, a depend of python2-pycparser, a depend of python2-cffi, a depend of python2-cryptography.

1:12 am

Updated python2-ply, python2-pycparser.

3:30 am

I had to put the device to charge. Updated python2-pycparser again, python2-cffi.

4:47 am

Updated python2-cryptography, and its dependency python2-ipaddress.

I have reported an issue in the latter to its maintainer with an aurweb comment.

5:08 am

Now you can python2-trustme.

I underline you can't make this joke with python3-trustme.

The remaining pytest deps are quite long to complete, so I'm taking a small pause.

4:20 pm

I wake up.

6:06 pm

I have chilled a couple hours on the telegram arch group. It's over 16 hours one of the devices is busy building nodejs to build python-sphinxcontrib-furo to have a working python-sphinx-theme-builder to build python-build.

For some reason I am not sure about, [main]'s python-build has a dependency loop on itself through its deps. I wonder if it wouldn't be better if it had a switch to self-build instead.

6:30 pm

I fix and reupload python2-pysocks. It has some missing checks so I will need to look at it again later.

8:20 pm

I'm currently lost in a labyrinth of dependencies, because python2-pysocks is related to python2-twisted, which still had lots of deps missing, as I knew already.

Anyway, I have just updated python2-twisted (proposed change, waiting for maintainer's reply), python2-priority, python2-soappy.

I am going to fry some frenches now.

Saturday 3th January 2024

I have fininhed upgraditg this a couple days ago. From when I have started I have updated around 100 new packages, mostly while rebuilding it.

The amount of checked multi-arch compatible packages resulting from having had to rebuild the python2 stack it's becoming significant, so I will probably publish a binary repo in a couple days.

Am I the only person thinking a package submitter (cough cough) saying 'I have an update, give me ownership to receive it' to its current, active maintainer instead of simply posting/sending it is kinda hostile?

I invite following -ur repositories development on Github before reporting out of dates or deletion/orphan requests to check ticket processing on my packages, you could be reporting something that's being pushed.

Merge/pull requests are welcomed.

The package works, in can be built with --nocheck.

Missing dependencies are being handled as in priority order on my schedule.

Missing dependencies can still be simply built from old commits syncing out upstream packages, it's quite easy, only time consuming to test if you are busy rebuilding main.

I am on a new arch so I don't have my 2020 onwards local repo available otherwise solving anything of this it would have been immediate.

I have been trying to build another package that is dependent on the above package (python2-urllib3).

Without the package python2-urllib3 I am unable to install acestream-engine.

Please remove all checkdepends, as it is now a wholly futile burden to build all of them with dependencies when none of those will get used in the end: the check() function is commented out due to testing having been unreliable for several users.

uninstalled it through pacman and installed it through "pip". there's no reason for it to be installed through pacman, when you can just install it through "pip", right?!

Builds for me if I deselect the failed tests in PKGBUILD file:

diff --git a/PKGBUILD b/PKGBUILD
index cb38f73..d7ff951 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -69,6 +69,10 @@ check() {
                --verbose \
                --cache-clear \
                --ignore test/contrib/test_socks.py \
+               --deselect test/contrib/test_pyopenssl.py::TestSSL::test_ssl_read_timeout \
+               --deselect test/with_dummyserver/test_connectionpool.py::TestConnectionPool::test_source_address \
+               --deselect test/with_dummyserver/test_connectionpool.py::TestConnectionPool::test_source_address_error \
+               --deselect test/with_dummyserver/test_socketlevel.py::TestSSL::test_ssl_read_timeout \
                --deselect test/test_retry.py::TestRetry::test_respect_retry_after_header_sleep \
                --deselect test/test_retry_deprecated.py::TestRetry::test_respect_retry_after_header_sleep \
                --deselect test/with_dummyserver/test_socketlevel.py::TestHeaders::test_request_host_header_ignores_fqdn_dot \

Don't know if this works reliably - but at least it builds and I can build further packages that depend on python2-urllib3 ^^

Test in Failed in line test_ssl_read_timeout FAILED and not install.