Openembedded Core Discussions
 help / color / mirror / Atom feed
From: "Yu, Mingli" <mingli.yu@windriver.com>
To: Matt Madison <matt@madison.systems>
Cc: Patches and discussions about the oe-core layer
	<openembedded-core@lists.openembedded.org>
Subject: Re: [PATCH] go-dep: disable PTEST_ENABLED
Date: Wed, 3 Jul 2019 10:34:09 +0800	[thread overview]
Message-ID: <5D1C1421.2070507@windriver.com> (raw)
In-Reply-To: <CAGgRHJqeFE75tXJ7qB4gyxkZaWsLmFozY+1YoJLUPXzG9Q+fOA@mail.gmail.com>



On 2019年07月01日 19:59, Matt Madison wrote:
> On Sun, Jun 30, 2019 at 6:59 PM Yu, Mingli <mingli.yu@windriver.com> wrote:
>>
>>
>>
>> On 2019年06月28日 19:02, Richard Purdie wrote:
>>> On Fri, 2019-06-28 at 00:57 -0700, mingli.yu@windriver.com wrote:
>>>> From: Mingli Yu <mingli.yu@windriver.com>
>>>>
>>>> The run-ptest logic for go-dep actually runs the
>>>> /usr/lib64/go-dep/ptest/github.com/golang/dep/cmd/dep/dep.test whose
>>>> source file is
>>>> https://github.com/golang/dep/blob/master/cmd/dep/dep_test.go.
>>>>
>>>> That dep_test.go starts by rebuilding the dep program
>>>> from source, then runs the tests using that copy of the
>>>> program, so it's assuming that we're still in a development
>>>> environment where we can run a full go build.
>>>>
>>>> Considering it not being designed for a cross-build setup,
>>>> so disable PTEST_ENABLED.
>>>>
>>>> Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
>>>> ---
>>>>    meta/recipes-devtools/go/go-dep_0.5.0.bb | 5 +++--
>>>>    1 file changed, 3 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/meta/recipes-devtools/go/go-dep_0.5.0.bb b/meta/recipes-
>>>> devtools/go/go-dep_0.5.0.bb
>>>> index a4d631f8ea..e9fc12fa5a 100644
>>>> --- a/meta/recipes-devtools/go/go-dep_0.5.0.bb
>>>> +++ b/meta/recipes-devtools/go/go-dep_0.5.0.bb
>>>> @@ -21,5 +21,6 @@ BBCLASSEXTEND = "native nativesdk"
>>>>
>>>>    # For compiling ptest on mips and mips64, the current go-dep version
>>>> fails with the go 1.11 toolchain.
>>>>    # error message: vet config not found
>>>> -PTEST_ENABLED_mips = "0"
>>>> -PTEST_ENABLED_mips64 = "0"
>>>> +# disable PTEST_ENABLED as the run-ptest script for go-dep actually
>>>> runs the /usr/lib64/go-
>>>> dep/ptest/github.com/golang/dep/cmd/dep/dep.test whose source file is
>>>> https://github.com/golang/dep/blob/master/cmd/dep/dep_test.go not
>>>> being designed for a cross-build setup.
>>>> +PTEST_ENABLED = "0"
>>>> +PTEST_ENABLED = "0"
>>>
>>> Setting it twice looks wrong.
>>
>> Sorry, it should be my typo.
>>
>>>
>>> If we're disabling it, why would we inherit the ptest class at all as
>>> its not going to work anywhere?
>>>
>>> Upstream not considering cross test usecases isn't a reason to disable
>>> a test, we have many tests enabled where upstream haven't considered a
>>> cross use case, we just tend to patch as needed and start a discussion
>>> with them.
>>>
>>> It sounds like its actually a network access problem from the image
>>> you're running into anyway?
>>
>> Hi RP,
>>
>> Have discussed the ptest more with Matt in the maillist and also tried
>> to add the patch under the guide from Matt to make the
>> https://github.com/golang/dep/blob/master/cmd/dep/dep_test.go work with
>> cross-setup env. But seems it still doesn't work.
>>
>> Hi Matt,
>> What's your opinion?
>
> I was able to get the test to work in my setup, with the addition of
> that patch and manually setting GOCACHE in my environment. As Richard

Thanks Matt's information!

If so, could you help to contribute your worked patch to oe-core?

I still got all cases failed even with the patch which to make 
https://github.com/golang/dep/blob/master/cmd/dep/dep_test.go 
correspondingly usable for cross-setup.

> mentioned, from the failures you were getting, it looked like a
> network access issue in your setup. I'm not sure why GOCACHE needs to
> be set, when 'go env' shows the correct (non-'off') default; that
> might be another issue in dep that needs to be addressed.
>
> It would be good to have at least one non-trivial ptest for a go
> package, and since OE-Core doesn't have much beyond the toolchain
> itself, go-dep, and glide, getting this to work would provide some
> much-needed coverage. Running go-dep or glide - both build tools - on
> a target seems like a fairly rare use case, though.

BTW, I noticed even go doesn't have ptest support.

>
> As for Richard's initial question, go.bbclass inherits ptest to enable
> the mapping of go's built-in test builds automatically. In hindsight,
> maybe this wasn't the best choice, and it was overly optimistic to
> think that because go supports separate compilation and execution of
> tests, that packages would generally support that by default.

How about to remove inherits ptest logic in go.bbclass and let the 
inherit ptest logic in each recipe if someone cares about go-dep ptest.

Thanks,

>
> -Matt
>
>> Thanks,
>>
>>>
>>> Cheers,
>>>
>>> Richard
>>>
>>>
>>>
>


  parent reply	other threads:[~2019-07-03  2:30 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-28  7:57 [PATCH] go-dep: disable PTEST_ENABLED mingli.yu
2019-06-28 11:02 ` Richard Purdie
2019-07-01  2:03   ` Yu, Mingli
2019-07-01 11:59     ` Matt Madison
2019-07-01 14:45       ` Burton, Ross
2019-07-03  2:34       ` Yu, Mingli [this message]
2019-07-03  9:24       ` [PATCH] go.bbclass: remove ptest inherit mingli.yu
2019-07-04 12:38         ` Matt Madison
2019-07-05  9:00           ` [PATCH v3] go.bbclass: separate the ptest logic to go-ptest class mingli.yu

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=5D1C1421.2070507@windriver.com \
    --to=mingli.yu@windriver.com \
    --cc=matt@madison.systems \
    --cc=openembedded-core@lists.openembedded.org \
    /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