[Bitbake-dev] what's with the "md5sum" wget fetcher parameter?
denis at denix.org
Sun Jan 17 04:02:06 CET 2010
On Sat, Jan 16, 2010 at 02:36:44PM -0700, Christopher Larson wrote:
> This is not correct. Robert is talking about the md5sum checking the
> bitbake fetch code can do based on a URL parameter, not the new method that
> Phil wrote in _OE_ which associates a name to the URL and stores the md5
> sum in a variable flag. Both checksums.ini and the bitbake fetch md5sum
> param checks are old methods, not current.
> -- Chris Larson
No, I was not talking about the name association in the URL to store md5 sum
in a variable flag! That's just a special case. The general case of the new
method is storing md5sum in the SRC_URI variable, which is handled and
implemented (and was for some time) in bitbake, not OE.
Storing md5 sums in the SRC_URIs of the corresponding recipes (vs. storing
them in the checksums.ini) is considered the new and recommended method not
because it was implemented recently (actually, it was already supported by
bitbake, so nothing needed to be implemented), but because we decided to
make it the default during OEDEM in November as a way to deprecate the use of
checksums.ini! It is explained in the first post from Phil I linked:
So, I was answering this Robert's question:
> note that, yes, the md5sum values match. but why have that info in
> both places? what's the standard these days?
And the answer is the same:
Using checksums.ini to store checksums was the old default method and is not
recommended any more. It is recommended now to store checksums in SRC_URIs of
the corresponding recipes. Is it more clear now?
> On Jan 16, 2010, at 1:07 PM, Denys Dmytriyenko <denis at denix.org> wrote:
>> On Sat, Jan 16, 2010 at 05:29:56AM -0500, Robert P. J. Day wrote:
>>> i note in the manual section on wget Fetcher the reference to the
>>> single supported "md5sum" parameter. so ...
>>> * it's not just that fetcher that supports that param, is it? i can
>>> see some md5sum-related methods defined in __init__.py -- doesn't that
>>> suggest that *all* Fetchers theoretically support that parameter?
>>> * more to the point, is anything actually checking those md5sum
>>> values? i thought the checksums.ini file was used for md5 and sha256
>>> so what happens if a .bb file has a SRC_URI directive that defines
>>> that parameter? is it processed? does it take precedence over the
>>> checksums file? thanks.
>> Please check the outcome of OEDEM'09 on the checksums:
>>> p.s. as an actual example, here's a snippet from checksums.ini:
>> This is the old way.
>>> and here's SRC_URI from recipes/net-tools/net-tools_1.60.bb:
>>> SRC_URI =
>> And this is the new preferred way.
>>> note that, yes, the md5sum values match. but why have that info in
>>> both places? what's the standard these days?
More information about the bitbake-devel