On Sun, 2021-08-22 at 14:50 +0200, Alexander Kanavin wrote:
> This lengthens dependency chains unnecessarily, and -native
> versions are already included in DEPENDS.
> 
> Signed-off-by: Alexander Kanavin <[email protected]>
> ---
>  meta/recipes-devtools/binutils/binutils_2.37.bb | 2 --
>  1 file changed, 2 deletions(-)
> 
> diff --git a/meta/recipes-devtools/binutils/binutils_2.37.bb 
> b/meta/recipes-devtools/binutils/binutils_2.37.bb
> index ff9377049b..e0fea913a3 100644
> --- a/meta/recipes-devtools/binutils/binutils_2.37.bb
> +++ b/meta/recipes-devtools/binutils/binutils_2.37.bb
> @@ -1,8 +1,6 @@
>  require binutils.inc
>  require binutils-${PV}.inc
>  
> -DEPENDS += "flex bison zlib"
> -

This is something which has caused a lot of problems in the past. In particular,
if something touches (patches( one of the pre-generated files or we use a git
version of binutils this breaks. It also seemed to depend on the host version of
patch used.

For zlib, is the recipe using the binary only or the library? If the latter,
that also proves problematic.

I appreciate it may have passed autobuilder testing but I suspect that isn't
enough for real work use.

Also, the target version would still need these?

I like the idea of streamlining dependencies but this one needs some further
checks...

Cheers,

Richard

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#155130): 
https://lists.openembedded.org/g/openembedded-core/message/155130
Mute This Topic: https://lists.openembedded.org/mt/85060380/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to