Re: [Dev] [PATCH abslibre] pacman: Get architecture from CARCH instead of auto

2016-08-24 Thread Paul Kocialkowski 
Le mardi 23 août 2016 à 12:55 -0500, Isaac David a écrit :
> Le mar. 23 août 2016 à 4:51, Paul Kocialkowski  a 
> écrit :
> > Also, I don't see what problem can arise from this, as it was 
> > suggested at:
> > https://labs.parabola.nu/issues/1039
> > 
> > ALARM also hardcodes "armv7h" in the distributed pacman.conf, so 
> > there's no
> > specific reason it would cause a problem when migrating. (Or am I 
> > perhaps
> > missing something here ?)
> 
> You should read it as:
> 
> "Using armv7l as the architecture name in pacman is the KISS solution,
>  but it would cause problems with migrating, etc."
> 
> instead of
> 
> "Hardcoding armv7h would cause problems with migrating, etc."

Ah right, makes more sense. Well, I don't think that using something different
from what uname -m returns as arch name is a problem. I think the hard-floating
point distinction is more relevant than endianness, because it restricts the
scope of possibly supported devices.

If anything, I think pacman's autodetection should be fixed to  return armv7h
when uname -m returns armv7l.

I could bring-in that change if you think it's relevant. Maybe upstream would be
interested too.

-- 
Paul Kocialkowski, developer of low-level free software for embedded devices

Website: https://www.paulk.fr/
Coding blog: https://code.paulk.fr/
Git repositories: https://git.paulk.fr/ https://git.code.paulk.fr/

signature.asc
Description: This is a digitally signed message part
___
Dev mailing list
Dev@lists.parabola.nu
https://lists.parabola.nu/mailman/listinfo/dev

Re: [Dev] [PATCH abslibre] pacman: Get architecture from CARCH instead of auto

2016-08-23 Thread Isaac David 
Le mar. 23 août 2016 à 4:51, Paul Kocialkowski  a 
écrit :


Also, I don't see what problem can arise from this, as it was 
suggested at:

https://labs.parabola.nu/issues/1039

ALARM also hardcodes "armv7h" in the distributed pacman.conf, so 
there's no
specific reason it would cause a problem when migrating. (Or am I 
perhaps

missing something here ?)


You should read it as:

"Using armv7l as the architecture name in pacman is the KISS solution,
but it would cause problems with migrating, etc."

instead of

"Hardcoding armv7h would cause problems with migrating, etc."

___
Dev mailing list
Dev@lists.parabola.nu
https://lists.parabola.nu/mailman/listinfo/dev

Re: [Dev] [PATCH abslibre] pacman: Get architecture from CARCH instead of auto

2016-08-23 Thread Paul Kocialkowski 
Hi,

Le vendredi 19 août 2016 à 20:20 -0500, Isaac David a écrit :
> This was bug https://labs.parabola.nu/issues/1039, thanks
> for reminding me of it.

Thanks for pushing a fix in the meantime.

> Were you able to test this configuration? Behind the scenes
> ALARM (should I say AGLARM?) is also doing some extra sed
> magic to substitute @CARCH@ with a hard-coded value at build
> time; so I can almost tell there's no way of using 'auto' or
> anything equivalent under AGLARM. Pacman does use the value
> of uname -m with 'auto' though,

Right, I hadn't dug enough to figure this out.

>  which raises the question:
> how did AGLARM come about giving the architecture this name?

The kernel uses the final letter "l" to indicate endianness. ALARM on the other
hand uses "h" to indicate hard floating point.

> The hack makes sense for them, because they chose to use a
> single pacman.conf template for all the architectures they
> maintain; but we have multiple files, like Arch does, in
> abslibre and therefore can afford just hard-coding "armv7h"
> in there. That's exactly what I did.

I think that's the right thing to do here.

Cheers,

-- 
Paul Kocialkowski, developer of low-level free software for embedded devices

Website: https://www.paulk.fr/
Coding blog: https://code.paulk.fr/
Git repositories: https://git.paulk.fr/ https://git.code.paulk.fr/

signature.asc
Description: This is a digitally signed message part
___
Dev mailing list
Dev@lists.parabola.nu
https://lists.parabola.nu/mailman/listinfo/dev

Re: [Dev] [PATCH abslibre] pacman: Get architecture from CARCH instead of auto

2016-08-19 Thread Isaac David 

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Hello Paul,

This was bug https://labs.parabola.nu/issues/1039, thanks
for reminding me of it.

Were you able to test this configuration? Behind the scenes
ALARM (should I say AGLARM?) is also doing some extra sed
magic to substitute @CARCH@ with a hard-coded value at build
time; so I can almost tell there's no way of using 'auto' or
anything equivalent under AGLARM. Pacman does use the value
of uname -m with 'auto' though, which raises the question:
how did AGLARM come about giving the architecture this name?

The hack makes sense for them, because they chose to use a
single pacman.conf template for all the architectures they
maintain; but we have multiple files, like Arch does, in
abslibre and therefore can afford just hard-coding "armv7h"
in there. That's exactly what I did.

All the best.

- --isacdaavid

-BEGIN PGP SIGNATURE-
Version: GnuPG v2
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=jHkV
-END PGP SIGNATURE-

___
Dev mailing list
Dev@lists.parabola.nu
https://lists.parabola.nu/mailman/listinfo/dev