[bitbake-devel] are the fetch-related variables in bitbake's bitbake.conf of any value?

[Paul Eggleton]

On Friday 20 June 2014 12:23:12 Robert P. J. Day wrote:
> On Fri, 20 Jun 2014, Christopher Larson wrote:
> > On Fri, Jun 20, 2014 at 9:00 AM, Robert P. J. Day <rpjday at crashcourse.ca>
> > wrote:
> > 
> >   more specific question related to my earlier post -- is there any
> > value whatever in the variables in bitbake's bitbake.conf file of the
> > form:
> > 
> >   FETCHCOMMAND*
> >   RESUMECOMMAND*
> >   UPDATECOMMAND*
> >   MKTEMP*CMD
> > 
> > nothing seems to use any of those. can they all just be deleted?
> > certainly, the FETCHCOMMAND* variables have been superseded by the
> > FETCHCMD* variables, no? i don't know about the remainder of them.
> > 
> > As far as I know, as you say, those were deprecated in favor of the
> > FETCHCMD vars, so can almost certainly be dropped.
> 
>   one more question before i submit a patch to do some cleaning.
> AFAICT, the bitbake.conf settings for RESUMECOMMAND* and
> UPDATECOMMAND* variables can be removed as well since nothing in
> bitbake refers to those variables, and it seems logical that bitbake
> should not be setting variables that it doesn't directly use in some
> way.

Those look old to me as well.

>   however, i'm puzzled by this from the bitbake codebase:
> 
> $ grep -r "MKTEMP.*CMD" *
> conf/bitbake.conf:MKTEMPCMD = "mktemp -q ${TMPBASE}"
> conf/bitbake.conf:MKTEMPDIRCMD = "mktemp -d -q ${TMPBASE}"
> lib/toaster/orm/fixtures/orm_views_testdata.json:        "variable_name":
> "MKTEMPCMD" lib/toaster/orm/fixtures/orm_views_testdata.json:       
> "variable_name": "MKTEMPDIRCMD" $
> 
>   so the only reference to the MKTEMP-related variables are in the
> toaster directory in some .json files. what does that mean? if one
> wanted to remove the MKTEMP* variables, would one also have to adjust
> those .json files accordingly?

The intention is to provide a way for QA verify that Toaster is collecting the 
variable data correctly, but because the json files contain every variable 
value, they are already out-of-date. We probably ought to filter the list that 
is used for that test to make it more maintainable.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre