[honister] docs: fix hardcoded link warning messages

From: Michael Opdenacker <michael.opdenacker@bootlin.com>

Sphinx complains about hardcoded links which can be replaced by an
extlink.

So let's apply its recommendations.

Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Reported-by: Quentin Schulz <foss+yocto@0leil.net>
---
 documentation/bsp-guide/bsp.rst                  | 7 ++++---
 documentation/migration-guides/migration-3.4.rst | 2 +-
 documentation/overview-manual/yp-intro.rst       | 4 ++--
 documentation/ref-manual/system-requirements.rst | 2 +-
 4 files changed, 8 insertions(+), 7 deletions(-)

Hi Michael,

On 3/22/22 15:25, Michael Opdenacker via lists.yoctoproject.org wrote:
> From: Michael Opdenacker <michael.opdenacker@bootlin.com>
> 
> Sphinx complains about hardcoded links which can be replaced by an
> extlink.
> 
> So let's apply its recommendations.
> 

If Sphinx is now happy,. I'm happy :)

BTW, you might want to send patches for honister tags to 
yocto-autobuilder-helper too otherwise we'll have build failures once we 
update sphinx package in the doc toolchain.

Thanks!
Quentin

Hi Quentin,

On 3/22/22 15:35, Quentin Schulz wrote:
> Hi Michael,
>
> On 3/22/22 15:25, Michael Opdenacker via lists.yoctoproject.org wrote:
>> From: Michael Opdenacker <michael.opdenacker@bootlin.com>
>>
>> Sphinx complains about hardcoded links which can be replaced by an
>> extlink.
>>
>> So let's apply its recommendations.
>>
>
> If Sphinx is now happy,. I'm happy :)


You were actually the first one to have a recent version of Sphinx and
propose such a patch for master.

>
> BTW, you might want to send patches for honister tags to
> yocto-autobuilder-helper too otherwise we'll have build failures once
> we update sphinx package in the doc toolchain.


... and for dunfell and hardknott tags too.
This sounds like a lot on not productive work, I'd prefer to contribute
to "master" instead :-|

I saw your latest patch
(https://git.yoctoproject.org/yocto-autobuilder-helper/commit/scripts/run-docs-build?id=8273124feb9da2ffe93fcee7c4529d5597e1684a)
and I understand that we don't break docs anymore when there is a
failure. But what about just ignoring warnings? At least these ones
don't have any impact on the generated docs.

Cheers
Michael.

Hi Michael,

On 3/22/22 15:58, Michael Opdenacker wrote:
> Hi Quentin,
> 
> On 3/22/22 15:35, Quentin Schulz wrote:
>> Hi Michael,
>>
>> On 3/22/22 15:25, Michael Opdenacker via lists.yoctoproject.org wrote:
>>> From: Michael Opdenacker <michael.opdenacker@bootlin.com>
>>>
>>> Sphinx complains about hardcoded links which can be replaced by an
>>> extlink.
>>>
>>> So let's apply its recommendations.
>>>
>>
>> If Sphinx is now happy,. I'm happy :)
> 
> 
> You were actually the first one to have a recent version of Sphinx and
> propose such a patch for master.
> 
>>
>> BTW, you might want to send patches for honister tags to
>> yocto-autobuilder-helper too otherwise we'll have build failures once
>> we update sphinx package in the doc toolchain.
> 
> 
> ... and for dunfell and hardknott tags too.

Actually no. Those releases don't have the -W flag set.

> This sounds like a lot on not productive work, I'd prefer to contribute
> to "master" instead :-|
>

There'll be work involved for each Sphinx recipe update probably.

> I saw your latest patch
> (https://urldefense.proofpoint.com/v2/url?u=https-3A__git.yoctoproject.org_yocto-2Dautobuilder-2Dhelper_commit_scripts_run-2Ddocs-2Dbuild-3Fid-3D8273124feb9da2ffe93fcee7c4529d5597e1684a&d=DwICaQ&c=_sEr5x9kUWhuk4_nFwjJtA&r=LYjLexDn7rXIzVmkNPvw5ymA1XTSqHGq8yBP6m6qZZ4njZguQhZhkI_-172IIy1t&m=7KvpXu-zWDEHrScQwpPEb1d7lQpqDn3tM0R6OX9vQ-setWP9vzh2eAaoa8Kh6YRC&s=jAUfJi_04ri8OvUg8fg9WTJCBYEqLd6QPSd08IqlW3c&e= )
> and I understand that we don't break docs anymore when there is a
> failure. But what about just ignoring warnings? At least these ones
> don't have any impact on the generated docs.
> 

Missing terms from bitbake are warnings IIRC and it's quite a big deal 
if we're not failing on those. I guess if we were able to mark only some 
warnings as errors, such as is possible with compilers for example, then 
I could imagine us doing something like this, yes. But in the 5s I 
looked for this I didn't find anything :/

Now that sphinx will be built by Yocto, we'll probably have much newer 
versions of Sphinx with more and more warnings (turned into errors). 
We'll need to see if it's worth patching or not or if we keep older 
toolchains (which kind of defeats the purpose of using Yocto Project to 
build the toolchain?).

Cheers,
Quentin