From: Jeremy Spewock <jspewock@iol.unh.edu>
To: "Juraj Linkeš" <juraj.linkes@pantheon.tech>
Cc: thomas@monjalon.net, Honnappa.Nagarahalli@arm.com,
probb@iol.unh.edu, paul.szczepanek@arm.com,
Luca.Vizzarro@arm.com, dev@dpdk.org
Subject: Re: [PATCH v2 6/7] dts: refactor logging configuration
Date: Wed, 14 Feb 2024 11:51:37 -0500 [thread overview]
Message-ID: <CAAA20UQYKCDi+0EtSXrqwZ3YS=htLLj3sN1gYuM+C=oY=D0piA@mail.gmail.com> (raw)
In-Reply-To: <CAOb5WZb-6_bF0OMbhh=7W+Gp7MNfHRNRupx07LpzmzpRC_wtEQ@mail.gmail.com>
On Wed, Feb 14, 2024 at 2:49 AM Juraj Linkeš <juraj.linkes@pantheon.tech> wrote:
>
> Hi Jeremy,
>
> Just a reminder, please strip the parts you're not commenting on.
>
<snip>
> > > + def __init__(self, *args, **kwargs):
> > > + """Extend the constructor with extra file handlers."""
> > > + self._extra_file_handlers = []
> > > + super().__init__(*args, **kwargs)
> > >
> > > - One handler logs to the console, the other one to a file, with either a regular or verbose
> > > - format.
> > > + def makeRecord(self, *args, **kwargs):
> >
> > Is the return type annotation here skipped because of the `:meta private:`?
> >
>
> Basically. We're modifying a method defined elsewhere, but there's no
> harm in adding the return type.
>
<snip>
> > > + logging.setLoggerClass(DTSLogger)
> > > + if name:
> > > + name = f"{dts_root_logger_name}.{name}"
> > > + else:
> > > + name = dts_root_logger_name
> > > + logger = logging.getLogger(name)
> > > + logging.setLoggerClass(logging.Logger)
> >
> > What's the benefit of setting the logger class back to logging.Logger
> > here? Is the idea basically that if someone wanted to use the logging
> > module we shouldn't implicitly make them use our DTSLogger?
> >
>
> Yes. We should actually set it to whatever was there before (it may
> not be logging.Logger), so I'll change that.+
>
>
<snip>
> > > @@ -137,6 +141,7 @@ def run(self):
> > >
> > > # for all Execution sections
> > > for execution in self._configuration.executions:
> > > + self._logger.set_stage(DtsStage.execution)
> >
> > This ends up getting set twice in short succession which of course
> > doesn't functionally cause a problem, but I don't exactly see the
> > point of setting it twice.
>
> I'm not sure what you're referring to. The stage is only set once at
> the start of each execution (and more generally, at the start of each
> stage). It's also set in each finally block to properly mark the
> cleanup of that stage. There are two finally blocks in the execution
> stage where that could happen, but otherwise it should be set exactly
> once.
You're right, I misread where it was being set the second time in the
code when I was reading this and thought there was a way where it
could get set twice but it cannot. Apologies, it makes sense that you
weren't sure what I was referring to because what I was referring to
doesn't exist.
>
> > We could either set it here or set it in
> > the _run_execution, but i think it makes more sense to set it in the
> > _run_execution method as that is the method where we are doing the
> > setup, here we are only initializing the nodes which is still in a
> > sense "pre execution."
>
> Init nodes was pre-execution before the patch. I've moved it to
> execution because of two reasons:
> 1. The nodes are actually defined per-execution. It just made more
> sense to move them to execution. Also, if an error happens while
> connecting to a node, we don't want to abort the whole DTS run, just
> the execution, as the next execution could be connecting to a
> different node.
> 2. We're not just doing node init here, we're also discovering which
> test cases to run. This is essential to do as soon as possible (before
> anything else happens in the execution, such as connecting the nodes)
> so that we can mark blocked test cases in case of an error. Test case
> discovery is definitely part of each execution and putting node init
> after that was a natural consequence. There's an argument for
> discovering test cases of all executions before actually running any
> of the executions. It's certainly doable, but the code doesn't look
> that good (when not modifying the original execution config (which we
> shouldn't do) - I've tried) and there's actually not much of a point
> to do it this way, the result is almost the same. Where it makes a
> difference is when there's an error in test suite configuration and
> later executions - the user would have to wait for the previous
> execution to fully run to discover the error).
>
These are very good points. I was still thinking of these
initializations as part of the pre-execution but I agree with you that
the things we are doing are actually part of the execution. Thank you
for elaborating!
> >
> >
<snip>
next prev parent reply other threads:[~2024-02-14 16:51 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-20 10:33 [RFC PATCH v1 0/5] test case blocking and logging Juraj Linkeš
2023-12-20 10:33 ` [RFC PATCH v1 1/5] dts: convert dts.py methods to class Juraj Linkeš
2023-12-20 10:33 ` [RFC PATCH v1 2/5] dts: move test suite execution logic to DTSRunner Juraj Linkeš
2023-12-20 10:33 ` [RFC PATCH v1 3/5] dts: process test suites at the beginning of run Juraj Linkeš
2023-12-20 10:33 ` [RFC PATCH v1 4/5] dts: block all testcases when earlier setup fails Juraj Linkeš
2023-12-20 10:33 ` [RFC PATCH v1 5/5] dts: refactor logging configuration Juraj Linkeš
2024-01-08 18:47 ` [RFC PATCH v1 0/5] test case blocking and logging Jeremy Spewock
2024-02-06 14:57 ` [PATCH v2 0/7] " Juraj Linkeš
2024-02-06 14:57 ` [PATCH v2 1/7] dts: convert dts.py methods to class Juraj Linkeš
2024-02-06 14:57 ` [PATCH v2 2/7] dts: move test suite execution logic to DTSRunner Juraj Linkeš
2024-02-06 14:57 ` [PATCH v2 3/7] dts: filter test suites in executions Juraj Linkeš
2024-02-12 16:44 ` Jeremy Spewock
2024-02-14 9:55 ` Juraj Linkeš
2024-02-06 14:57 ` [PATCH v2 4/7] dts: reorganize test result Juraj Linkeš
2024-02-06 14:57 ` [PATCH v2 5/7] dts: block all test cases when earlier setup fails Juraj Linkeš
2024-02-06 14:57 ` [PATCH v2 6/7] dts: refactor logging configuration Juraj Linkeš
2024-02-12 16:45 ` Jeremy Spewock
2024-02-14 7:49 ` Juraj Linkeš
2024-02-14 16:51 ` Jeremy Spewock [this message]
2024-02-06 14:57 ` [PATCH v2 7/7] dts: improve test suite and case filtering Juraj Linkeš
2024-02-23 7:54 ` [PATCH v3 0/7] test case blocking and logging Juraj Linkeš
2024-02-23 7:54 ` [PATCH v3 1/7] dts: convert dts.py methods to class Juraj Linkeš
2024-02-23 7:54 ` [PATCH v3 2/7] dts: move test suite execution logic to DTSRunner Juraj Linkeš
2024-02-23 7:54 ` [PATCH v3 3/7] dts: filter test suites in executions Juraj Linkeš
2024-02-27 21:21 ` Jeremy Spewock
2024-02-28 9:16 ` Juraj Linkeš
2024-02-23 7:54 ` [PATCH v3 4/7] dts: reorganize test result Juraj Linkeš
2024-02-23 7:55 ` [PATCH v3 5/7] dts: block all test cases when earlier setup fails Juraj Linkeš
2024-02-23 7:55 ` [PATCH v3 6/7] dts: refactor logging configuration Juraj Linkeš
2024-02-23 7:55 ` [PATCH v3 7/7] dts: improve test suite and case filtering Juraj Linkeš
2024-02-27 21:24 ` [PATCH v3 0/7] test case blocking and logging Jeremy Spewock
2024-03-01 10:55 ` [PATCH v4 " Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 1/7] dts: convert dts.py methods to class Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 2/7] dts: move test suite execution logic to DTSRunner Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 3/7] dts: filter test suites in executions Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 4/7] dts: reorganize test result Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 5/7] dts: block all test cases when earlier setup fails Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 6/7] dts: refactor logging configuration Juraj Linkeš
2024-03-01 10:55 ` [PATCH v4 7/7] dts: improve test suite and case filtering Juraj Linkeš
2024-03-01 17:41 ` Jeremy Spewock
2024-03-01 17:43 ` Jeremy Spewock
2024-03-07 11:04 ` Thomas Monjalon
2024-03-01 16:11 ` [PATCH v4 0/7] test case blocking and logging Patrick Robb
2024-03-07 14:55 ` Thomas Monjalon
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAAA20UQYKCDi+0EtSXrqwZ3YS=htLLj3sN1gYuM+C=oY=D0piA@mail.gmail.com' \
--to=jspewock@iol.unh.edu \
--cc=Honnappa.Nagarahalli@arm.com \
--cc=Luca.Vizzarro@arm.com \
--cc=dev@dpdk.org \
--cc=juraj.linkes@pantheon.tech \
--cc=paul.szczepanek@arm.com \
--cc=probb@iol.unh.edu \
--cc=thomas@monjalon.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).