bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Ludovic Courtès]

Gash-Utils ‘uname’ seems to be confused:

--8<---------------cut here---------------start------------->8---
$ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -a
Linux ribbon 5.10.10-gnu #1 SMP 1 x86_64 GNU/Linux
$ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -s

$ guix describe
Generacio 175	Feb 04 2021 22:52:40	(nuna)
  guix 5ae09d7
    repository URL: https://git.savannah.gnu.org/git/guix.git
    branch: master
    commit: 5ae09d7979a0696d862b9555314eab199f7ce576
$ uname -s
Linux
--8<---------------cut here---------------end--------------->8---

Ludo’.

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Efraim Flashner]

[-- Attachment #1.1: Type: text/plain, Size: 1109 bytes --]

On Mon, Feb 15, 2021 at 06:24:46PM +0100, Ludovic Courtès wrote:
> Gash-Utils ‘uname’ seems to be confused:
> 
> --8<---------------cut here---------------start------------->8---
> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -a
> Linux ribbon 5.10.10-gnu #1 SMP 1 x86_64 GNU/Linux
> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -s
> 
> $ guix describe
> Generacio 175	Feb 04 2021 22:52:40	(nuna)
>   guix 5ae09d7
>     repository URL: https://git.savannah.gnu.org/git/guix.git
>     branch: master
>     commit: 5ae09d7979a0696d862b9555314eab199f7ce576
> $ uname -s
> Linux
> --8<---------------cut here---------------end--------------->8---
> 
> Ludo’.
> 

The attached patch fixes it for me when I build it locally. Tested
against master, with an expanded guix.scm to build multiple versions.

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #1.2: 0001-uname-Fix-uname-s.patch --]
[-- Type: text/plain, Size: 996 bytes --]

From eedd0ee769393eea3fcb9b7131d469ea15d9f179 Mon Sep 17 00:00:00 2001
From: Efraim Flashner <efraim@flashner.co.il>
Date: Tue, 16 Feb 2021 11:50:20 +0200
Subject: [PATCH] uname: Fix uname -s.

* gash/commands/uname.scm (uname): Remove duplicate kernel-name? match.
---
 gash/commands/uname.scm | 1 -
 1 file changed, 1 deletion(-)

diff --git a/gash/commands/uname.scm b/gash/commands/uname.scm
index 2a89c6a..fa15c69 100644
--- a/gash/commands/uname.scm
+++ b/gash/commands/uname.scm
@@ -49,7 +49,6 @@
          (processor? (option-ref options 'processor #f))
          (hardware-platform? (option-ref options 'hardware-platform #f))
          (operating-system? (option-ref options 'operating-system #f))
-         (kernel-name? (not (or all? kernel-name? node-name? kernel-release? kernel-version? machine? processor? hardware-platform? operating-system?)))
 
 	 (help? (option-ref options 'help #f))
          (version? (option-ref options 'version #f))
-- 
2.30.1


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Ludovic Courtès]

Hi,

(+Cc: gash-devel.)

Efraim Flashner <efraim@flashner.co.il> skribis:

> On Mon, Feb 15, 2021 at 06:24:46PM +0100, Ludovic Courtès wrote:
>> Gash-Utils ‘uname’ seems to be confused:
>> 
>> --8<---------------cut here---------------start------------->8---
>> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -a
>> Linux ribbon 5.10.10-gnu #1 SMP 1 x86_64 GNU/Linux
>> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -s
>> 
>> $ guix describe
>> Generacio 175	Feb 04 2021 22:52:40	(nuna)
>>   guix 5ae09d7
>>     repository URL: https://git.savannah.gnu.org/git/guix.git
>>     branch: master
>>     commit: 5ae09d7979a0696d862b9555314eab199f7ce576
>> $ uname -s
>> Linux
>> --8<---------------cut here---------------end--------------->8---

I forgot to mention the ‘uname -s’ problem is hidden by the fact that we
always pass ‘--build’ to ‘configure’.  When we don’t do that,
‘config.guess’ runs and fails.

> From eedd0ee769393eea3fcb9b7131d469ea15d9f179 Mon Sep 17 00:00:00 2001
> From: Efraim Flashner <efraim@flashner.co.il>
> Date: Tue, 16 Feb 2021 11:50:20 +0200
> Subject: [PATCH] uname: Fix uname -s.
>
> * gash/commands/uname.scm (uname): Remove duplicate kernel-name? match.
> ---
>  gash/commands/uname.scm | 1 -
>  1 file changed, 1 deletion(-)
>
> diff --git a/gash/commands/uname.scm b/gash/commands/uname.scm
> index 2a89c6a..fa15c69 100644
> --- a/gash/commands/uname.scm
> +++ b/gash/commands/uname.scm
> @@ -49,7 +49,6 @@
>           (processor? (option-ref options 'processor #f))
>           (hardware-platform? (option-ref options 'hardware-platform #f))
>           (operating-system? (option-ref options 'operating-system #f))
> -         (kernel-name? (not (or all? kernel-name? node-name? kernel-release? kernel-version? machine? processor? hardware-platform? operating-system?)))

I wonder if this should be kept but changed to:

  (kernel-name? (or kernel-name? (not (or …))))

but I’m not entirely sure about the logic.

Gash folks, WDYT?

Thanks,
Ludo’.

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 2709 bytes --]

On Wed, Feb 17, 2021 at 04:16:29PM +0100, Ludovic Courtès wrote:
> Hi,
> 
> (+Cc: gash-devel.)
> 
> Efraim Flashner <efraim@flashner.co.il> skribis:
> 
> > On Mon, Feb 15, 2021 at 06:24:46PM +0100, Ludovic Courtès wrote:
> >> Gash-Utils ‘uname’ seems to be confused:
> >> 
> >> --8<---------------cut here---------------start------------->8---
> >> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -a
> >> Linux ribbon 5.10.10-gnu #1 SMP 1 x86_64 GNU/Linux
> >> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -s
> >> 
> >> $ guix describe
> >> Generacio 175	Feb 04 2021 22:52:40	(nuna)
> >>   guix 5ae09d7
> >>     repository URL: https://git.savannah.gnu.org/git/guix.git
> >>     branch: master
> >>     commit: 5ae09d7979a0696d862b9555314eab199f7ce576
> >> $ uname -s
> >> Linux
> >> --8<---------------cut here---------------end--------------->8---
> 
> I forgot to mention the ‘uname -s’ problem is hidden by the fact that we
> always pass ‘--build’ to ‘configure’.  When we don’t do that,
> ‘config.guess’ runs and fails.
> 
> > From eedd0ee769393eea3fcb9b7131d469ea15d9f179 Mon Sep 17 00:00:00 2001
> > From: Efraim Flashner <efraim@flashner.co.il>
> > Date: Tue, 16 Feb 2021 11:50:20 +0200
> > Subject: [PATCH] uname: Fix uname -s.
> >
> > * gash/commands/uname.scm (uname): Remove duplicate kernel-name? match.
> > ---
> >  gash/commands/uname.scm | 1 -
> >  1 file changed, 1 deletion(-)
> >
> > diff --git a/gash/commands/uname.scm b/gash/commands/uname.scm
> > index 2a89c6a..fa15c69 100644
> > --- a/gash/commands/uname.scm
> > +++ b/gash/commands/uname.scm
> > @@ -49,7 +49,6 @@
> >           (processor? (option-ref options 'processor #f))
> >           (hardware-platform? (option-ref options 'hardware-platform #f))
> >           (operating-system? (option-ref options 'operating-system #f))
> > -         (kernel-name? (not (or all? kernel-name? node-name? kernel-release? kernel-version? machine? processor? hardware-platform? operating-system?)))
> 
> I wonder if this should be kept but changed to:
> 
>   (kernel-name? (or kernel-name? (not (or …))))

With it changed to this:
(ins)efraim@3900XT ~/workspace/gash-utils$ /gnu/store/9mb28yyzfh6pr4yk5nyy42vrfbj8caxp-gash-utils-boot-0.1.0.107-eedd-dirty/bin/uname -s
Linux

> 
> but I’m not entirely sure about the logic.
> 
> Gash folks, WDYT?
> 
> Thanks,
> Ludo’.


-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Timothy Sample]

Hello!

Efraim Flashner <efraim@flashner.co.il> writes:

> On Wed, Feb 17, 2021 at 04:16:29PM +0100, Ludovic Courtès wrote:
>>
>> Efraim Flashner <efraim@flashner.co.il> skribis:
>> 
>> > On Mon, Feb 15, 2021 at 06:24:46PM +0100, Ludovic Courtès wrote:
>> >> Gash-Utils ‘uname’ seems to be confused:
>> >> 
>> >> --8<---------------cut here---------------start------------->8---
>> >> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -a
>> >> Linux ribbon 5.10.10-gnu #1 SMP 1 x86_64 GNU/Linux
>> >> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -s
>> >> 
>> >> $ guix describe
>> >> Generacio 175	Feb 04 2021 22:52:40	(nuna)
>> >>   guix 5ae09d7
>> >>     repository URL: https://git.savannah.gnu.org/git/guix.git
>> >>     branch: master
>> >>     commit: 5ae09d7979a0696d862b9555314eab199f7ce576
>> >> $ uname -s
>> >> Linux
>> >> --8<---------------cut here---------------end--------------->8---
>> 
>> I forgot to mention the ‘uname -s’ problem is hidden by the fact that we
>> always pass ‘--build’ to ‘configure’.  When we don’t do that,
>> ‘config.guess’ runs and fails.
>> 
>> [...]
>> 
>> I wonder if this should be kept but changed to:
>> 
>>   (kernel-name? (or kernel-name? (not (or …))))
>
> With it changed to this:
> (ins)efraim@3900XT ~/workspace/gash-utils$
> /gnu/store/9mb28yyzfh6pr4yk5nyy42vrfbj8caxp-gash-utils-boot-0.1.0.107-eedd-dirty/bin/uname
> -s
> Linux

Indeed!  I noticed this same issue the other day.  Check out commit
2ee2e2721367cc242ceb0e4b364468870a87f0b4:

    https://git.savannah.nongnu.org/cgit/gash/gash-utils.git/commit/?id=2ee2e2721367cc242ceb0e4b364468870a87f0b4

I recommend that you take a look at the “wip-gcc-mesboot” branch.  It
has a couple “from the hip” changes that I made while looking at getting
Gash-Utils to handle more of the work in “commencement.scm”.  Last week
I was able to build to “gcc-mesboot1” with only Make, Diffutils, patch,
and Gzip (the first three were there before the “Scheme-only” bootstrap,
and Gzip saves writing “(delete 'compress-documentation)” over and
over).  Unfortunately, building Glibc after that is quite difficult.  To
do it with Gash-Utils, we would need (at least) to support redirects in
AWK and the “-t” and “-k” flags in sort.  For fun, I tried building Gawk
and Coreutils at this point (using the ancient versions that we
bootstrap with now).  That let me build Glibc, but then I hit a problem
with the “po2test.sed” script while building “gcc-mesboot”.  I tried
again with ancient GNU Sed, but the build seemed to hang during a
configure test.  I suspect adding Bash to the mix might fix it, but I
haven’t tried yet....

Back to the matter at hand: I’m closing this bug since it’s already been
fixed.  :)


-- Tim

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Ludovic Courtès]

Hi!

Timothy Sample <samplet@ngyro.com> skribis:

> Efraim Flashner <efraim@flashner.co.il> writes:
>
>> On Wed, Feb 17, 2021 at 04:16:29PM +0100, Ludovic Courtès wrote:
>>>
>>> Efraim Flashner <efraim@flashner.co.il> skribis:
>>> 
>>> > On Mon, Feb 15, 2021 at 06:24:46PM +0100, Ludovic Courtès wrote:
>>> >> Gash-Utils ‘uname’ seems to be confused:
>>> >> 
>>> >> --8<---------------cut here---------------start------------->8---
>>> >> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -a
>>> >> Linux ribbon 5.10.10-gnu #1 SMP 1 x86_64 GNU/Linux
>>> >> $ $(guix build -e '(@@ (gnu packages commencement) gash-utils-boot)')/bin/uname -s
>>> >> 
>>> >> $ guix describe
>>> >> Generacio 175	Feb 04 2021 22:52:40	(nuna)
>>> >>   guix 5ae09d7
>>> >>     repository URL: https://git.savannah.gnu.org/git/guix.git
>>> >>     branch: master
>>> >>     commit: 5ae09d7979a0696d862b9555314eab199f7ce576

[...]

> Indeed!  I noticed this same issue the other day.  Check out commit
> 2ee2e2721367cc242ceb0e4b364468870a87f0b4:
>
>     https://git.savannah.nongnu.org/cgit/gash/gash-utils.git/commit/?id=2ee2e2721367cc242ceb0e4b364468870a87f0b4

Great!

> I recommend that you take a look at the “wip-gcc-mesboot” branch.  It
> has a couple “from the hip” changes that I made while looking at getting
> Gash-Utils to handle more of the work in “commencement.scm”.  Last week
> I was able to build to “gcc-mesboot1” with only Make, Diffutils, patch,
> and Gzip (the first three were there before the “Scheme-only” bootstrap,
> and Gzip saves writing “(delete 'compress-documentation)” over and
> over).

We should add a pure Scheme gzip implementation, for instance using the
R6RS zlib implementation by Göran Weinholt in Industria.

> Unfortunately, building Glibc after that is quite difficult.  To do it
> with Gash-Utils, we would need (at least) to support redirects in AWK
> and the “-t” and “-k” flags in sort.  For fun, I tried building Gawk
> and Coreutils at this point (using the ancient versions that we
> bootstrap with now).  That let me build Glibc, but then I hit a
> problem with the “po2test.sed” script while building “gcc-mesboot”.  I
> tried again with ancient GNU Sed, but the build seemed to hang during
> a configure test.  I suspect adding Bash to the mix might fix it, but
> I haven’t tried yet....

Heheh, nice!  Anyhow, that looks like a promising path to me.

> Back to the matter at hand: I’m closing this bug since it’s already been
> fixed.  :)

Thanks!  Any plan for a release?  We could upgrade the package in
‘core-updates’.

Ludo’.

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Timothy Sample]

Hey,

Ludovic Courtès <ludo@gnu.org> writes:

> Timothy Sample <samplet@ngyro.com> skribis:
>
>> [...] Gzip saves writing “(delete 'compress-documentation)” over and
>> over).
>
> We should add a pure Scheme gzip implementation, for instance using the
> R6RS zlib implementation by Göran Weinholt in Industria.

Bootar is already using that.  The problem with the Industria version is
that it only does decompression.  Fortunately, there’s an easy out here.
Gzip supports an uncompressed mode, so it would take only an hour or so
to write a “compressor” that just copies the data into Gzip format with
no compression.  (I’m hoping that XZ can do something similar so that
applying patches in “origin” records does not need to depend on the XZ
we use to unpack Guile.  Alternatively, we could add an option to
compress the patched sources with Gzip.)

>> Unfortunately, building Glibc after that is quite difficult.  To do it
>> with Gash-Utils, we would need (at least) to support redirects in AWK
>> and the “-t” and “-k” flags in sort.  For fun, I tried building Gawk
>> and Coreutils at this point (using the ancient versions that we
>> bootstrap with now).  That let me build Glibc, but then I hit a
>> problem with the “po2test.sed” script while building “gcc-mesboot”.  I
>> tried again with ancient GNU Sed, but the build seemed to hang during
>> a configure test.  I suspect adding Bash to the mix might fix it, but
>> I haven’t tried yet....
>
> Heheh, nice!  Anyhow, that looks like a promising path to me.
>
>> Back to the matter at hand: I’m closing this bug since it’s already been
>> fixed.  :)
>
> Thanks!  Any plan for a release?  We could upgrade the package in
> ‘core-updates’.

I’ve been holding on to “gcc-mesboot” as the milestone that would
warrant a release.  On the other hand, I could try and reify the
progress so far into an improved “commencement.scm” that still uses some
old GNU utilities to build the last “mesboot” version of Glibc.  Either
way, it will take some time since (as you know) my attention is on
Disarchive right now.  :)


-- Tim

bug#46537: ‘uname -s’ from Gash-Utils-Boot returns nothing

[Ludovic Courtès]

Hi!

Timothy Sample <samplet@ngyro.com> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> Timothy Sample <samplet@ngyro.com> skribis:
>>
>>> [...] Gzip saves writing “(delete 'compress-documentation)” over and
>>> over).
>>
>> We should add a pure Scheme gzip implementation, for instance using the
>> R6RS zlib implementation by Göran Weinholt in Industria.
>
> Bootar is already using that.  The problem with the Industria version is
> that it only does decompression.  Fortunately, there’s an easy out here.
> Gzip supports an uncompressed mode, so it would take only an hour or so
> to write a “compressor” that just copies the data into Gzip format with
> no compression.  (I’m hoping that XZ can do something similar so that
> applying patches in “origin” records does not need to depend on the XZ
> we use to unpack Guile.  Alternatively, we could add an option to
> compress the patched sources with Gzip.)

Oh I see.  And yes, we could add an option to turn off compression or
force use of gzip.

>> Thanks!  Any plan for a release?  We could upgrade the package in
>> ‘core-updates’.
>
> I’ve been holding on to “gcc-mesboot” as the milestone that would
> warrant a release.  On the other hand, I could try and reify the
> progress so far into an improved “commencement.scm” that still uses some
> old GNU utilities to build the last “mesboot” version of Glibc.  Either
> way, it will take some time since (as you know) my attention is on
> Disarchive right now.  :)

Heheh, that’s great too.  :-)

Thanks!

Ludo’.