~alpine/devel

3 2

Why did pipeline build two packages?

Details
Message ID
<Y0Cq5VKmcRgkezYu@ws>
DKIM signature
missing
Download raw message
Hi,

I'm curious about this[0] job that was executed as part of pipeline for my
merge request. It looks like it tried to build two packages:

    >>> testing/bazel: build succesfully
    >>> testing/bazel4: build succesfully

Any idea why? My merge request[1] is touching just one of them, so I
have no idea why both are built. It seems a bit unnecessary and the
build takes quite long.

Given that I do not touch bazel4, I guess following error is expected?

    >>> bazel4: Updating the testing/x86_64 repository index...
    >>> bazel4: Signing the index...
    >>> ERROR: the built package (bazel4) is already in the repo

Any ideas?

Thank,
W.

0: https://gitlab.alpinelinux.org/gray-wolf/aports/-/jobs/865337
1: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/39933

-- 
There are only two hard things in Computer Science:
cache invalidation, naming things and off-by-one errors.
Details
Message ID
<CNG2LTDA05SC.3EVTYAI1YLLIF@sumire>
In-Reply-To
<Y0Cq5VKmcRgkezYu@ws> (view parent)
DKIM signature
missing
Download raw message
On Sat Oct 8, 2022 at 12:40 AM CEST, Wolf wrote:
> Hi,
>
> I'm curious about this[0] job that was executed as part of pipeline for my
> merge request. It looks like it tried to build two packages:
>
>     >>> testing/bazel: build succesfully
>     >>> testing/bazel4: build succesfully
>
> Any idea why? My merge request[1] is touching just one of them, so I
> have no idea why both are built. It seems a bit unnecessary and the
> build takes quite long.
>
> Given that I do not touch bazel4, I guess following error is expected?
>
>     >>> bazel4: Updating the testing/x86_64 repository index...
>     >>> bazel4: Signing the index...
>     >>> ERROR: the built package (bazel4) is already in the repo
>
> Any ideas?
probably because bazel4 provides=bazel and the thing that triggers the
rebuilds for CI gets confused by it. the new one should be named bazel5
in any case :)
>
> Thank,
> W.
>
> 0: https://gitlab.alpinelinux.org/gray-wolf/aports/-/jobs/865337
> 1: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/39933
>
> -- 
> There are only two hard things in Computer Science:
> cache invalidation, naming things and off-by-one errors.
Details
Message ID
<Y0HhHVza644D2ZhP@ws>
In-Reply-To
<CNG2LTDA05SC.3EVTYAI1YLLIF@sumire> (view parent)
DKIM signature
missing
Download raw message
On 2022-10-08 01:13:21 +0200, alice wrote:
> On Sat Oct 8, 2022 at 12:40 AM CEST, Wolf wrote:
> > Hi,
> >
> > I'm curious about this[0] job that was executed as part of pipeline for my
> > merge request. It looks like it tried to build two packages:
> >
> >     >>> testing/bazel: build succesfully
> >     >>> testing/bazel4: build succesfully
> >
> > Any idea why? My merge request[1] is touching just one of them, so I
> > have no idea why both are built. It seems a bit unnecessary and the
> > build takes quite long.
> >
> > Given that I do not touch bazel4, I guess following error is expected?
> >
> >     >>> bazel4: Updating the testing/x86_64 repository index...
> >     >>> bazel4: Signing the index...
> >     >>> ERROR: the built package (bazel4) is already in the repo
> >
> > Any ideas?
> probably because bazel4 provides=bazel and the thing that triggers the
> rebuilds for CI gets confused by it. the new one should be named bazel5
> in any case :)

I see, I did not notice that. I will add provides=bazel into my package
as well.

Sure, I will rename it, no problem. Is there any rule on whether
packages should include their version in the package name? For example
neither php nor nodejs do it (and there multiple versions would be
actually useful), but for example lua does. I assume there is no
clear-cut rule?

I will rename it of course, I don't really care about the name :)

W.

-- 
There are only two hard things in Computer Science:
cache invalidation, naming things and off-by-one errors.
Details
Message ID
<CNGXZEK1U52Q.39MX1JI7DVSQ@sumire>
In-Reply-To
<Y0HhHVza644D2ZhP@ws> (view parent)
DKIM signature
missing
Download raw message
On Sat Oct 8, 2022 at 10:44 PM CEST, Wolf wrote:
> On 2022-10-08 01:13:21 +0200, alice wrote:
> > On Sat Oct 8, 2022 at 12:40 AM CEST, Wolf wrote:
> > > Hi,
> > >
> > > I'm curious about this[0] job that was executed as part of pipeline for my
> > > merge request. It looks like it tried to build two packages:
> > >
> > >     >>> testing/bazel: build succesfully
> > >     >>> testing/bazel4: build succesfully
> > >
> > > Any idea why? My merge request[1] is touching just one of them, so I
> > > have no idea why both are built. It seems a bit unnecessary and the
> > > build takes quite long.
> > >
> > > Given that I do not touch bazel4, I guess following error is expected?
> > >
> > >     >>> bazel4: Updating the testing/x86_64 repository index...
> > >     >>> bazel4: Signing the index...
> > >     >>> ERROR: the built package (bazel4) is already in the repo
> > >
> > > Any ideas?
> > probably because bazel4 provides=bazel and the thing that triggers the
> > rebuilds for CI gets confused by it. the new one should be named bazel5
> > in any case :)
>
> I see, I did not notice that. I will add provides=bazel into my package
> as well.
>
> Sure, I will rename it, no problem. Is there any rule on whether
> packages should include their version in the package name? For example
> neither php nor nodejs do it (and there multiple versions would be
> actually useful), but for example lua does. I assume there is no
> clear-cut rule?
php does (7, 8, 81). nodejs would too if it had more than lts/current, i
guess..
so, only really the things with multiple versions for real (and not
simply "latest"/"lts"), but there's no clear cut rule.
>
> I will rename it of course, I don't really care about the name :)
>
> W.
>
> -- 
> There are only two hard things in Computer Science:
> cache invalidation, naming things and off-by-one errors.
Reply to thread Export thread (mbox)