SRC_URI checksums

SRC_URI checksums

[Darren Hart]

What is the practice for SRC_URI checksums? I see many recipes with both
md5sum and sha256sum. Is there a need to have both? Is one preferred
over the other?

Thanks,

-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Technical Lead - Linux Kernel

Re: SRC_URI checksums

[Jerrod Peach]

[-- Attachment #1: Type: text/plain, Size: 1249 bytes --]

On Wed, Jan 23, 2013 at 11:11 AM, Darren Hart <dvhart@linux.intel.com>wrote:

> What is the practice for SRC_URI checksums? I see many recipes with both
> md5sum and sha256sum. Is there a need to have both? Is one preferred
> over the other?
>
> Thanks,
>
> --
> Darren Hart
> Intel Open Source Technology Center
> Yocto Project - Technical Lead - Linux Kernel
> _______________________________________________
> yocto mailing list
> yocto@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto
>

I believe bitbake/Yocto will complain at you if you don't have both, but
I've always wondered this myself.  Since it's just an integrity check, I'd
think an md5sum would be sufficient and marginally faster than a sha256,
but if you wanted to be ultra paranoid, a sha256 would probably be good
enough.  The level of paranoia it would take to want to make both checks is
beyond what just about anyone would want, I'd suspect.  I'll append to the
original e-mailer's question: Should the default be to require both checks?
 Is there a simple way to turn off complaining about one check or the other
if you're not interested in running both?  Are both even run?  (I think
they both are...)

Kind regards,

Jerrod

[-- Attachment #2: Type: text/html, Size: 2016 bytes --]

SRC URI checksums

[Edward Wingate]

I have a recipe that has a SRC_URI retrieving from a mercurial repo:
SRC_URI = "hg://server/project;module=name;rev=tip"
SRC_URI[md5sum] = "xxx"
SRC_URI[sha256sum] = "yyy"

I changed the SRC_URL to retreive from a specific revision instead of tip:
SRC_URI = "hg://server/project;module=name;rev=1234567890"

I went ahead and did a bitbake, expecting it to complain about the
SRC_URI checksums not matching, and I would go ahead and update the
checksums from the error output.

But it succeeded, building with the correct revision.  I thought the
SRC_URI checksums had to be updated when the SRC_URI changes?

Ed