[gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[Kent Fredric]

[-- Attachment #1: Type: text/plain, Size: 1512 bytes --]

One of the recurring problems we face in #gentoo is end users coming
to us with confusing problems, and their problems are exacerbated
because their default workflow ended up with them unmasking some **
version of perl.

There is already a bug for this behaviour [1], and comments say that
portage doing this is "a bug", but the situations which it occurs in
are hard to diagnose what the "real problem" is.

Much of the time, what has occurred is there was some other problem,
and portage bodged its way around the real problem by choosing a
solution that should be considered unacceptable, instead of presenting
the real problem.

Some of the time, the cause is as simple as a single package being
installed that isn't in the @world dependency graph any more, which is
tripping up portage slot-rebuild behaviour.

In practice, what this currently means is that stable users end up
installing *developmental/experimental* packages that exist only for
experts and gentoo maintainers, and this is an unacceptable resolution.

If this behaviour was being triggered by anything other than portage's
dependency resolver failing, it would be considered a serious QA
violation.

Its understood that portage maintainers want to "fix" this behaviour so
the problem doesn't occur, but until that can be done, the present
default behaviour is actively harmful, and I suggest it be disabled by
default until it can be guaranteed to give the right results.

1: https://bugs.gentoo.org/658648

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[Zac Medico]

[-- Attachment #1.1: Type: text/plain, Size: 1758 bytes --]

On 10/9/19 4:53 PM, Kent Fredric wrote:
> One of the recurring problems we face in #gentoo is end users coming
> to us with confusing problems, and their problems are exacerbated
> because their default workflow ended up with them unmasking some **
> version of perl.
> 
> There is already a bug for this behaviour [1], and comments say that
> portage doing this is "a bug", but the situations which it occurs in
> are hard to diagnose what the "real problem" is.
> 
> Much of the time, what has occurred is there was some other problem,
> and portage bodged its way around the real problem by choosing a
> solution that should be considered unacceptable, instead of presenting
> the real problem.
> 
> Some of the time, the cause is as simple as a single package being
> installed that isn't in the @world dependency graph any more, which is
> tripping up portage slot-rebuild behaviour.
> 
> In practice, what this currently means is that stable users end up
> installing *developmental/experimental* packages that exist only for
> experts and gentoo maintainers, and this is an unacceptable resolution.
> 
> If this behaviour was being triggered by anything other than portage's
> dependency resolver failing, it would be considered a serious QA
> violation.
> 
> Its understood that portage maintainers want to "fix" this behaviour so
> the problem doesn't occur, but until that can be done, the present
> default behaviour is actively harmful, and I suggest it be disabled by
> default until it can be guaranteed to give the right results.
> 
> 1: https://bugs.gentoo.org/658648
> 

I'd prefer to disable --autounmask by default and include warnings about
harmful behavior in the documentation.
-- 
Thanks,
Zac


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 981 bytes --]

Re: [gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[Kent Fredric]

[-- Attachment #1: Type: text/plain, Size: 509 bytes --]

On Wed, 9 Oct 2019 19:45:34 -0700
Zac Medico <zmedico@gentoo.org> wrote:

> I'd prefer to disable --autounmask by default and include warnings about
> harmful behavior in the documentation.

I think autounmasks behaviour with regards to USE flags is useful,
(heh), its just everything else it does tends to violate stability
assumptions.

For instance, I can see why automatically escalating "arch" to "~arch"
may be useful in some cases, but its a very dangerous default,
*especially* for perl.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[Michael Everitt]

[-- Attachment #1.1: Type: text/plain, Size: 643 bytes --]

On 10/10/19 03:57, Kent Fredric wrote:
> On Wed, 9 Oct 2019 19:45:34 -0700
> Zac Medico <zmedico@gentoo.org> wrote:
>
>> I'd prefer to disable --autounmask by default and include warnings about
>> harmful behavior in the documentation.
> I think autounmasks behaviour with regards to USE flags is useful,
> (heh), its just everything else it does tends to violate stability
> assumptions.
>
> For instance, I can see why automatically escalating "arch" to "~arch"
> may be useful in some cases, but its a very dangerous default,
> *especially* for perl.
Is there any mileage in something as ridiculous as AUTOUNMASK_EXCLUDES ?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 801 bytes --]

Re: [gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[Kent Fredric]

[-- Attachment #1: Type: text/plain, Size: 636 bytes --]

On Thu, 10 Oct 2019 04:35:22 +0100
Michael Everitt <gentoo@veremit.xyz> wrote:

> Is there any mileage in something as ridiculous as AUTOUNMASK_EXCLUDES ?

Would depend on what that means, where that variable was used, and what
it contains....

But the sort of breakage type it can trigger is not limited to any
specific package.

Pretty much any ** packages are prone to this happening, and that's so
unacceptable, that people have suggested/considered removing **
packages just to avoid this issue.

And that pretty much makes the defined purpose of those packages
useless, and makes that keywording grade useless.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-dev] [RFC] Disable --autounmask auto-unmasking keywords / masks by default

[Zac Medico]

[-- Attachment #1.1: Type: text/plain, Size: 851 bytes --]

On 10/9/19 7:57 PM, Kent Fredric wrote:
> On Wed, 9 Oct 2019 19:45:34 -0700
> Zac Medico <zmedico@gentoo.org> wrote:
> 
>> I'd prefer to disable --autounmask by default and include warnings about
>> harmful behavior in the documentation.
> 
> I think autounmasks behaviour with regards to USE flags is useful,
> (heh), its just everything else it does tends to violate stability
> assumptions.
> 
> For instance, I can see why automatically escalating "arch" to "~arch"
> may be useful in some cases, but its a very dangerous default,
> *especially* for perl.

We can add --autounmask-use=<y|n> option that's enabled by default when
--autounmask is unspecified.

When --autounmask=y is specified, we'll have full autounmask behavior.

When --autounmask=n is specified, we'll have --autounmask-use=n behavior.
-- 
Thanks,
Zac


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 981 bytes --]