RE: Custom recipe, multiple machines for same arch

[Daniel Machon <dama@universal-robots.com>]

[-- Attachment #1: Type: text/plain, Size: 2073 bytes --]



> -----Original Message-----
> From: Jan Kiszka <jan.kiszka@siemens.com>
> Sent: 3. marts 2021 13:05
> To: Daniel Machon <dama@universal-robots.com>; isar-
> users@googlegroups.com
> Subject: Re: Custom recipe, multiple machines for same arch
>
>
> CAUTION - EXTERNAL EMAIL: Do not open attachments or links unless you
> recognize the sender and the content is safe.
>
>
> On 03.03.21 12:37, Daniel Machon wrote:
> > Hi,
> >
> >
> >
> > Is there an elegant way to handle custom recipes, that are used in
> > multiple machines that target the same architecture.
> >
> > Currently when I try to execute a multi-machine build, it fails with:
> > */Detect multiple executions of ../*
> >
> > You can get around this by changing PN = /recipe-name/-${MACHINE} in
> > each custom recipe and IMAGE_INSTALL += /recipe-name/-${MACHINE}
> >
> > I am just wondering whether this is the best solution, or if there are
> > any alternatives?
> >
>
> Yes, you can share recipes across multiple configuration (multi-config).
> You will find examples of that in Isar itself and also in [1].
>
> But this can easily cause false-sharing problems. To detect those, we installed an
> instrumentation that reported such a case in your setup. It basically detects if
> there are variables evaluated by the recipe that vary across the builds for
> different multi-config targets.
>
> The next thing then is to understand what makes the recipes specific to the
> machines and then resolve that. One way is building the recipes machine-specific,
> this is what you described above. But maybe there is also a more elegant way in
> your case. This is hard to answer generally, though. You may try to compare the
> task signatures of the different runs via bitbake-diffsigs which should visualize the
> varying vars.
>
> Jan
>
> --
> Siemens AG, T RDA IOT
> Corporate Competence Center Embedded Linux
[Daniel Machon]

I will stick with the machine-specific recipes for now - this seems to be a reasonable solution in this case.
Thank you 😊


[-- Attachment #2: Type: text/html, Size: 2963 bytes --]