Default SRCREV value

Default SRCREV value

[Mikhail Gusarov]

[-- Attachment #1: Type: text/plain, Size: 279 bytes --]

Hello,

I've stumbled upon SRCREV="1" setting if bitbake.conf today, debugging
misterious mis-build of pkg-native.

Shouldn't SRCREV default to ${AUTOREV} or something like
${sys.write("Please define SRCREV"}; sys.exit(1)} ? Default value of "1"
is hardly useful.

-- 

[-- Attachment #2: Type: application/pgp-signature, Size: 196 bytes --]

Re: Default SRCREV value

[Philip Balister]

[-- Attachment #1: Type: text/plain, Size: 844 bytes --]

Mikhail Gusarov wrote:
> Hello,
> 
> I've stumbled upon SRCREV="1" setting if bitbake.conf today, debugging
> misterious mis-build of pkg-native.
> 
> Shouldn't SRCREV default to ${AUTOREV} or something like
> ${sys.write("Please define SRCREV"}; sys.exit(1)} ? Default value of "1"
> is hardly useful.

We know :)

But defaulting to ${AUTOREV} causes builds to fail if the repository is 
offline during parsing and checking AUTOREV during parsing is a slow 
process.

Setting SRCREV to 1 is not a good solution, but it is better than all 
the alternatives. Personally, I wish bitbake would print a HUGE message 
when it tries to fetch a SRCREV of 1 telling people to update a 
preferred version file or add AUTOREV to a local conf file. (I have 
AUTOREV in my local.conf for a few bb files where I mean AUTOREV)

Philip

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 3303 bytes --]

Re: Default SRCREV value

[Holger Freyther]

On Thursday 20 March 2008 00:50:35 Philip Balister wrote:

> Setting SRCREV to 1 is not a good solution, but it is better than all
> the alternatives. Personally, I wish bitbake would print a HUGE message
> when it tries to fetch a SRCREV of 1 telling people to update a
> preferred version file or add AUTOREV to a local conf file.

Or as Mikhail suggested:
${@bb.fatal("You want to specify a sane srvrev")}

The only down-side I can think of, now parsing will fail for a recipe you do 
not even care for in your distribution, with having a default value 1 you 
just parse fine.


z.

Re: Default SRCREV value

[Philip Balister]

[-- Attachment #1: Type: text/plain, Size: 934 bytes --]

Hmm, I am asleep at the wheel today. Good suggeestion!

Philip

Holger Freyther wrote:
> On Thursday 20 March 2008 00:50:35 Philip Balister wrote:
> 
>> Setting SRCREV to 1 is not a good solution, but it is better than all
>> the alternatives. Personally, I wish bitbake would print a HUGE message
>> when it tries to fetch a SRCREV of 1 telling people to update a
>> preferred version file or add AUTOREV to a local conf file.
> 
> Or as Mikhail suggested:
> ${@bb.fatal("You want to specify a sane srvrev")}
> 
> The only down-side I can think of, now parsing will fail for a recipe you do 
> not even care for in your distribution, with having a default value 1 you 
> just parse fine.
> 
> 
> z.
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> 

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 3303 bytes --]

Re: Default SRCREV value

[Mikhail Gusarov]

[-- Attachment #1: Type: text/plain, Size: 519 bytes --]

Twas brillig at 01:20:47 20.03.2008 UTC+01 when Holger Freyther did gyre and gimble:

 HF> The only down-side I can think of, now parsing will fail for a
 HF> recipe you do not even care for in your distribution, with having a
 HF> default value 1 you just parse fine.

Is it possible to handle recipies a bit lazier, to avoid recipies which
are not interesting?

Hmm... In other hand, until recipie is not parsed, it is not possible to
say whether it is needed for build or not, so it can be tricky.

-- 

[-- Attachment #2: Type: application/pgp-signature, Size: 196 bytes --]

Re: Default SRCREV value

[Leon Woestenberg]

Hello,

On Thu, Mar 20, 2008 at 1:20 AM, Holger Freyther <zecke@selfish.org> wrote:
>  Or as Mikhail suggested:
>  ${@bb.fatal("You want to specify a sane srvrev")}
>
>  The only down-side I can think of, now parsing will fail for a recipe you do
>  not even care for in your distribution, with having a default value 1 you
>  just parse fine.
>
I am +1 for this, it's another step to more robustness IMHO.

(And if you don't care with the version, then just live with a sane
revision *we* chose).

Regards,
-- 
Leon