Richard Purdie <[email protected]> writes:
>> Instead of adding immediate variable name expansion, the current syntax
>> could be changed to still do late/delayed variable name expansion, but
>> let the last assignment win. By giving each assignment an incremental
>> assignment-number, contest between multiple variables names could be
>> resolved by the latest assignmnet. Not sure exactly what would be the
>> natural way to handle appends and prepends in such a scenario though.
>
> This would be totally unlike any behaviour we currently have. Wouldn't
> that just add more confusion to the mix?
How is that totally unlike any behaviour we currently have?
I really believe that most people would expect
BAR = "bar"
FOO_${BAR} = "1"
FOO_bar = "2"
to result in FOO_bar to be "2". Of-course, those with enough OE
experience might rightly know that this will produce FOO_bar to be "1",
and even know why.
But as
FOO_bar = "1"
FOO_bar = "2"
gives FOO_bar "2", it is not _that_ unlike anything current, if you ask
me.
Anyways, it is clearly a change, and will likely have some metadata
impact, if nothing else, then perhaps in some of the
FILES_${PN}-.... variables, if someone injected the expanded versions of
the variables in some recipe (which is currently being ignored).
/Esben
_______________________________________________
Bitbake-dev mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/bitbake-dev
