Rocksolid Light

News from da outaworlds

mail  files  register  groups  login

Message-ID:  

You will contract a rare disease.


comp / comp.unix.bsd.freebsd.misc / Re: [ports] Shouldn't we rebuild all dependant packages whenever a library changes?

SubjectAuthor
* [ports] Shouldn't we rebuild all dependant packages whenever a library changes?Robin Haberkorn
`* Re: [ports] Shouldn't we rebuild all dependant packages whenever a library changJohn Levine
 `- Re: [ports] Shouldn't we rebuild all dependant packages whenever a library changRobin Haberkorn

1
Subject: [ports] Shouldn't we rebuild all dependant packages whenever a library changes?
From: Robin Haberkorn
Newsgroups: comp.unix.bsd.freebsd.misc
Organization: A noiseless patient Spider
Date: Sat, 19 Oct 2024 19:27 UTC
Path: eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: robin.haberkorn@googlemail.com (Robin Haberkorn)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: [ports] Shouldn't we rebuild all dependant packages whenever a
library changes?
Date: Sat, 19 Oct 2024 22:27:13 +0300
Organization: A noiseless patient Spider
Lines: 24
Message-ID: <8a5d1c33-906e-f61d-d39a-ad5a560fc24f@googlemail.com>
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset=US-ASCII
Injection-Date: Sat, 19 Oct 2024 21:27:17 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="5d26962ba79257eb6156deb9a44914cf";
logging-data="24273"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX18Jo7edp+sfFwiRWIJ7C/RTe/aXXV1TlnY="
User-Agent: Alpine 2.26 (BSF 649 2022-06-02)
Cancel-Lock: sha1:3/s0Yc32IDiVEFjV7M7PVGJ9k/s=
View all headers

Hello!

I stumbled across a strange glitch and not for the first time. Did a pkg
upgrade, which updated Qt (new minor version). This breaks all Qt apps
installed via pkg since they were built against a different minor version
of Qt, but were not rebuilt and pushed to the repository along with Qt.
Qt checks the library versions on startup, so you notice this immediately.
The current behavior (or lack of behavior) however could affect __all__
applications, even those that link only against plain C libraries. It's
rare, but you can't exclude that some minor library release breaks the API
- ABI is of course not a problem in C, compared to C++.

Workaround is to (re)build the corresponding application manually from the
ports tree. This can take hours, so IMHO this is a serious issue.
It is NOT sufficient to reinstall the package in question - it obviously
didn't change in the repository.

I would therefore think that every library change should result in a
rebuild of all dependant packages and automatic bumping of the package version.

Why is this not done? And where should we file a bug report?

Best regards,
Robin

Subject: Re: [ports] Shouldn't we rebuild all dependant packages whenever a library changes?
From: John Levine
Newsgroups: comp.unix.bsd.freebsd.misc
Organization: Taughannock Networks
Date: Sat, 19 Oct 2024 20:49 UTC
References: 1
Path: eternal-september.org!news.eternal-september.org!feeder3.eternal-september.org!panix!weretis.net!feeder9.news.weretis.net!news.misty.com!news.iecc.com!.POSTED.news.iecc.com!not-for-mail
From: johnl@taugh.com (John Levine)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: [ports] Shouldn't we rebuild all dependant packages whenever a
library changes?
Date: Sat, 19 Oct 2024 20:49:10 -0000 (UTC)
Organization: Taughannock Networks
Message-ID: <vf1606$2v1o$1@gal.iecc.com>
References: <8a5d1c33-906e-f61d-d39a-ad5a560fc24f@googlemail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Injection-Date: Sat, 19 Oct 2024 20:49:10 -0000 (UTC)
Injection-Info: gal.iecc.com; posting-host="news.iecc.com:2001:470:1f07:1126:0:676f:7373:6970";
logging-data="97336"; mail-complaints-to="abuse@iecc.com"
In-Reply-To: <8a5d1c33-906e-f61d-d39a-ad5a560fc24f@googlemail.com>
Cleverness: some
X-Newsreader: trn 4.0-test77 (Sep 1, 2010)
Originator: johnl@iecc.com (John Levine)
View all headers

It appears that Robin Haberkorn <robin.haberkorn@googlemail.com> said:
>applications, even those that link only against plain C libraries. It's
>rare, but you can't exclude that some minor library release breaks the API
>- ABI is of course not a problem in C, compared to C++.

The difference between a minor and major version is that minor versions have
the same interfaces. If the interfaces change, thaat's a major version.

It sounds like the bug is that a breaking change was shipped as a minor version
rather than a major one.

--
Regards,
John Levine, johnl@taugh.com, Primary Perpetrator of "The Internet for Dummies",
Please consider the environment before reading this e-mail. https://jl.ly

Subject: Re: [ports] Shouldn't we rebuild all dependant packages whenever a library changes?
From: Robin Haberkorn
Newsgroups: comp.unix.bsd.freebsd.misc
Organization: A noiseless patient Spider
Date: Mon, 21 Oct 2024 17:06 UTC
References: 1 2
Path: eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: robin.haberkorn@googlemail.com (Robin Haberkorn)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: [ports] Shouldn't we rebuild all dependant packages whenever a
library changes?
Date: Mon, 21 Oct 2024 20:06:43 +0300
Organization: A noiseless patient Spider
Lines: 21
Message-ID: <c09c9091-acd9-820b-eff0-4703cf60215f@googlemail.com>
References: <8a5d1c33-906e-f61d-d39a-ad5a560fc24f@googlemail.com> <vf1606$2v1o$1@gal.iecc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Injection-Date: Mon, 21 Oct 2024 19:06:47 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="30eceef714edf33d8a22462128bb0785";
logging-data="1094045"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX1+smNCGA5S3s2SPom60tKArfhRQTHDxBJY="
User-Agent: Alpine 2.26 (BSF 649 2022-06-02)
Cancel-Lock: sha1:MmRgVPQbFN3orKrtBszcyZRzz+s=
In-Reply-To: <vf1606$2v1o$1@gal.iecc.com>
View all headers

On Sat, 19 Oct 2024, John Levine wrote:

> The difference between a minor and major version is that minor versions have
> the same interfaces. If the interfaces change, thaat's a major version.
>
That's how it should be, but since the version comes from upstream, you
can't be sure about it.

> It sounds like the bug is that a breaking change was shipped as a minor version
> rather than a major one.
>

I am no longer that sure about it.
I recompiled my VirtualBox port and it did __not__ resolve the issue.
Then I did `pkg install --force qt5-style-plugins`.
That did the job. Obviously the package already was automatically rebuilt,
but the version was not bumped, so pkg upgrade wouldn't pick it up
automatically. If the rebuild of VirtualBox really was necessary - who
knows. At least force-reinstalling it via pkg wasn't enough.

There definitely is some kind of bug.

1

rocksolid light 0.9.8
clearnet tor