[PATCH core-updates-frozen 0/52] Support cross-compilation in glib-or-gtk-build-system and fix cross-compilation errors

Hi guix,

This patch series does two things:

* let 'glib-or-gtk-build-system' support cross-compilation

* fix many (but not all) cross-compilation errors that prevent

"guix build gtk+ --target=aarch64-linux-gnu" from succeeding.

It should not cause any rebuilds. You can test with

./pre-inst-env guix build gtk+ --target=aarch64-linux-gnu.

There is still some work to be done though:

cannot build derivation `[...]-cups-filters-1.28.9.drv': 1 dependencies couldn't be built

[...] `[...]-cups-2.3.3op2.drv': 1 dependencies couldn't be built

[...] `[...]-samba-4.13.10.drv': 6 dependencies couldn't be built

[...] `[...]-libsoup-2.72.0.drv': 4 dependencies couldn't be built

[...] `[...]-rest-0.8.1.drv': 1 dependencies couldn't be built

Greetings,

Maxime

This allows cross-compilation to succeed when not using

QEMU's transparent emulation, and the cross-compilation

is not a pointless cross-compilation like x86_64 -> i686.

to a native version of "dbus-binding-tool" when cross-compiling.

---

gnu/packages/glib.scm | 10 ++++++++++

1 file changed, 10 insertions(+)

%build-inputs and friends don't exist when cross-compiling,

so use some G-expology instead.

Don't use '%outputs' or '%build-inputs' when cross-compiling.

---

gnu/packages/freedesktop.scm | 14 +++++++++++---

1 file changed, 11 insertions(+), 3 deletions(-)

(elogind)[arguments]{configure-flags}: Let this be a G-exp.

---

gnu/packages/freedesktop.scm | 56 +++++++++++++++++++-----------------

1 file changed, 29 insertions(+), 27 deletions(-)

(libcap)[arguments]<#:phases>{configure}: Don't use non-existent

'%output' when cross-compiling.

(libcap)[arguments]<#:make-flags>: Set "CROSS_COMPILE" and "BUILD_CC"

when cross-compiling.

---

gnu/packages/linux.scm | 38 +++++++++++++++++++++++++-------------

1 file changed, 25 insertions(+), 13 deletions(-)

(nspr)[arguments]<#:configure-flags>: Set --host and --target

inappropriately when cross-compiling.

---

gnu/packages/nss.scm | 20 +++++++++++++++-----

1 file changed, 15 insertions(+), 5 deletions(-)

(perl)[arguments]<#:phases>{unpack-cross}: Search for "bash" with

'search-input-file' instead of 'assoc-ref'.

(perl)[arguments]<#:phases>{configure}: Search for "bash-minimal"

instead of "bash".

---

gnu/packages/perl.scm | 9 +++++----

1 file changed, 5 insertions(+), 4 deletions(-)

(nspr)[inputs]: Add 'perl' and 'bash-minimal' when cross-compiling.

(nspr)[arguments]{#:disallowed-references}: Add native 'perl' when

cross-compiling.

---

gnu/packages/nss.scm | 18 +++++++++++++++++-

1 file changed, 17 insertions(+), 1 deletion(-)

(poppler)[arguments]<#:phases>{set-PKG_CONFIG}: New phase.

---

gnu/packages/pdf.scm | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)

Presumably the other versions of mozjs need some changes

as well, but for now only fix mozjs@60, which is a dependency

of polkit.

(mozjs-60)[arguments]<#:configure-flags>: Set --with-nspr-prefix

and set --host and --target inappropriately when cross-compiling.

(mozjs-60)[arguments]<#:phases>{set-PKG_CONFIG}: New phase

when cross-compiling.

---

gnu/packages/gnuzilla.scm | 51 ++++++++++++++++++++++++++++-----------

1 file changed, 37 insertions(+), 14 deletions(-)

(meson-build, meson-cross-build): Only call 'sexp->gexp' on

'configure-flags' when 'configure-flags' is a pair.

---

guix/build-system/meson.scm | 9 +++++++--

1 file changed, 7 insertions(+), 2 deletions(-)

If absent, this causes the 'glib-or-gtk-wrap' phase to fail.

(cairo)[inputs]: Add "bash-minimal", but only when cross-compiling

to avoid a world-rebuild.

---

gnu/packages/gtk.scm | 7 ++++++-

1 file changed, 6 insertions(+), 1 deletion(-)

Alternative ‘os types’ include RedHat and Gentoo, which

don't seem applicable.

Set "--with-os-type" to "unknown".

---

gnu/packages/polkit.scm | 7 ++++++-

1 file changed, 6 insertions(+), 1 deletion(-)

(lower): Add 'implicit-cross-inputs?' argument. Generate a bag

when cross-compiling.

(glib-or-gtk-cross-build): New procedure.

---

guix/build-system/glib-or-gtk.scm | 146 ++++++++++++++++++++++++++----

1 file changed, 127 insertions(+), 19 deletions(-)

(gobject-introspection)[arguments]<#:configure-flags>: Set

gi_cross_use_prebuilt_gi to 'true'.

(gobject-introspection)[arguments]<#:phases>{set-typelibs}: New phase,

avoid undefined variable in "meson.build".

---

gnu/packages/glib.scm | 22 +++++++++++++++++++++-

1 file changed, 21 insertions(+), 1 deletion(-)

(gobject-introspection)[native-inputs]: Add 'bison' and 'flex'

when cross-compiling.

(gobject-introspection)[inputs]: Use 'python' instead of 'python-wrapper'

when cross-compiling. Remove 'bison', 'flex' when cross-compiling.

---

gnu/packages/glib.scm | 16 ++++++++++++----

1 file changed, 12 insertions(+), 4 deletions(-)

(gobject-introspection)[arguments]<#:phases>{rename-library}:

Correct the name of the python extension when cross-compiling.

---

gnu/packages/glib.scm | 37 ++++++++++++++++++++++++++++---------

1 file changed, 28 insertions(+), 9 deletions(-)

(atk)[arguments]<#:configure-flags>: Set -Dintrospection=false

when cross-compiling.

---

gnu/packages/gtk.scm | 9 ++++++++-

1 file changed, 8 insertions(+), 1 deletion(-)

introspection=false when cross-compiling.

---

gnu/packages/gtk.scm | 6 +++++-

1 file changed, 5 insertions(+), 1 deletion(-)

(bdb-4.8)[arguments]<#:phases>{update-config.sub}: Update 'config.sub'

when cross-compiling for 'powerpc64le' or 'aarch64'.

---

gnu/packages/dbm.scm | 21 +++++++++++++++++++++

1 file changed, 21 insertions(+)

(gamin)[arguments]<#:phases>{replace-config.sub}: Update 'config.sub'

when cross-compiling.

(gamin)[native-inputs]: Add "config" when cross-compiling.

---

gnu/packages/gnome.scm | 13 ++++++++++++-

1 file changed, 12 insertions(+), 1 deletion(-)

(iso-codes)[inputs]: Move "gettext", "perl" and "python" to ...

(iso-codes)[native-inputs]: ... here when cross-compiling.

---

gnu/packages/iso-codes.scm | 15 ++++++++++++---

1 file changed, 12 insertions(+), 3 deletions(-)

(heimdal)[inputs]: Add "bash-minimal" when cross-compiling.

(heimdal)[native-inputs]: Add 'perl' when cross-compiling.

---

gnu/packages/kerberos.scm | 11 ++++++++++-

1 file changed, 10 insertions(+), 1 deletion(-)

(gamin)[arguments]<#:phases>{set-have-abstract-sockets}: Override

test for ‘abstract sockets’ when cross-compiling.

---

gnu/packages/gnome.scm | 26 +++++++++++++++++++++++++-

1 file changed, 25 insertions(+), 1 deletion(-)

(heimdal)[arguments]<#:phases>{pre-configure}: Let "appl/afsutil/pagsh.c"

refer to the cross-compiled bash instead of the native bash when

cross-compiling.

---

gnu/packages/kerberos.scm | 23 ++++++++++++++++++-----

1 file changed, 18 insertions(+), 5 deletions(-)

Set ac_cv_func_getpwnam_r_posix=yes when cross-compiling.

---

gnu/packages/kerberos.scm | 9 +++++++--

1 file changed, 7 insertions(+), 2 deletions(-)

(heimdal)[arguments]<#:phases>{pre-configure}: Replace _PATH_BSHELL

with the cross-compiled bin/sh when cross-compiling.

---

gnu/packages/kerberos.scm | 10 ++++++++--

1 file changed, 8 insertions(+), 2 deletions(-)

(heimdal)[arguments]<#:phases>{pre-configure}: Consider ${krb_cv_com_err}

to be 'yes' when cross-compiling.

---

gnu/packages/kerberos.scm | 4 ++++

1 file changed, 4 insertions(+)

(heimdal)[arguments]<#:configure-flags>: Set --with-cross-tools.

(heimdal)[arguments]<#:phases>{pre-configure}: Override

ac_cv_PROG_COMPILE_ET.

---

gnu/packages/kerberos.scm | 51 ++++++++++++++++++++++++---------------

1 file changed, 32 insertions(+), 19 deletions(-)

cross-compiling (for wrapper scripts).

---

gnu/packages/gnome.scm | 5 +++++

1 file changed, 5 insertions(+)

Look up "docbook-xml" in 'native-inputs', not 'inputs', when

cross-compiling.

---

gnu/packages/gnome.scm | 15 ++++++++++++---

1 file changed, 12 insertions(+), 3 deletions(-)

(at-spi2-core)[arguments]<#:phases>{patch-docbook-sgml}: Look up

"docbook-sxml" in 'native-inputs' instead of 'inputs' when cross-compiling.

---

gnu/packages/gtk.scm | 13 ++++++++++---

1 file changed, 10 insertions(+), 3 deletions(-)

(at-spi2-core)[inputs]: Add "bash-minimal" when cross-compiling.

---

gnu/packages/gtk.scm | 5 +++++

1 file changed, 5 insertions(+)

(json-glib)[arguments]<#:configue-flags>: Set gtk_doc=disables when

cross-compiling.

(json-glib)[arguments]<#:phases>{move-docs}: Don't run when cross-compiling

...

(json-glib)[arguments]<#:phases>{stub-docs}: ... instead, create an empty

directory where the documentation would be.

---

gnu/packages/gnome.scm | 31 +++++++++++++++++++++----------

1 file changed, 21 insertions(+), 10 deletions(-)

(libxinerama)[arguments]<#:configure-flags>: Add malloc0-flags.

---

gnu/packages/xorg.scm | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)

The idea is to keep the comments in one place, instead of copying

them over and over. It's also documented more now.

(malloc0-flags): New procedure.

(libxext,libxrender,libx11,libxt): Use new procedures.

---

gnu/packages/xorg.scm | 32 ++++++++++++++------------------

1 file changed, 14 insertions(+), 18 deletions(-)

(libxxf86vm)[arguments]<#:configure-flags>: Add malloc0-flags.

---

gnu/packages/xorg.scm | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)

(libxi)[arguments]<#:configure-flags>: Add malloc0-flags.

---

gnu/packages/xorg.scm | 5 +++--

1 file changed, 3 insertions(+), 2 deletions(-)

(wayland)[arguments]<#:phases>{patchdocbook-xml}: Look in 'native-inputs'

instead of 'inputs' when cross-compiling.

---

gnu/packages/freedesktop.scm | 14 +++++++++++---

1 file changed, 11 insertions(+), 3 deletions(-)

(gdk-pixbuf)[inputs]: Add "bash-minimal" when cross-compiling.

---

gnu/packages/gtk.scm | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)

(at-spi2-atk)[arguments]<#:configure-flags>: Set tests=false when

cross-compiling.

---

gnu/packages/gtk.scm | 4 ++++

1 file changed, 4 insertions(+)

pkg-config-for-build and 'this-package' when cross-compiling.

---

gnu/packages/freedesktop.scm | 6 +++++-

1 file changed, 5 insertions(+), 1 deletion(-)

(libproxy)[arguments]]<#:phases>{check}: Respect #:tests? when

cross-compiling.

---

gnu/packages/networking.scm | 10 ++++++++--

1 file changed, 8 insertions(+), 2 deletions(-)

This allows the cross-build to succeed.

(libthai)[native-inputs]: Add "datrie" when cross-compiling.

---

gnu/packages/gtk.scm | 6 +++++-

1 file changed, 5 insertions(+), 1 deletion(-)

(libxrandr)[arguments]<#:configure-flags>: Add malloc0-flags.

---

gnu/packages/xorg.scm | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)

(at-spi2-core)[arguments]<#:configure-flags>: Set docs=false when

cross-compiling.

(at-spi2-core)[arguments]<#:phases>{move-documentation}: Remove when

cross-compiling.

---

gnu/packages/gtk.scm | 30 ++++++++++++++++++------------

1 file changed, 18 insertions(+), 12 deletions(-)

(pango)[inputs]: Add "bash-minimal" when cross-compiling.

---

gnu/packages/gtk.scm | 6 +++++-

1 file changed, 5 insertions(+), 1 deletion(-)

The configure script first tests whether "pkg-config" is in PATH,

and then uses "TARGET-pkg-config". Pretend "pkg-config" exists.

ac_cv_prog_have_pkg_config=yes when cross-compiling.

---

gnu/packages/avahi.scm | 8 ++++++--

1 file changed, 6 insertions(+), 2 deletions(-)

(json-glib)[arguments]<#:configure-flags>: Set introspection=false when

cross-compiling.

---

gnu/packages/gnome.scm | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)

Replace the bash in the shebangs in "/etc/avahi" with a cross-compiled bash.

---

gnu/packages/avahi.scm | 25 +++++++++++++++++++++++--

1 file changed, 23 insertions(+), 2 deletions(-)

(gdk-pixbuf)[arguments]<#:phases>{patch-docbook}: Look for docbook

in 'native-inputs' instead of 'inputs' when cross-compiling.

---

gnu/packages/gtk.scm | 14 +++++++++++---

1 file changed, 11 insertions(+), 3 deletions(-)

(pkg-config-for-build): New variable.

---

gnu/packages/pkg-config.scm | 36 ++++++++++++++++++++++++++++++++++++

1 file changed, 36 insertions(+)

(libdaemon)[native-inputs]: Add "config" when cross-compiling for

aarch64.

(libdaemon)[arguments]<#:phases>{update-config.sub}: When cross-compiling

for aarch64, update the "config.sub" script.

---

gnu/packages/libdaemon.scm | 42 ++++++++++++++++++++++++++------------

1 file changed, 29 insertions(+), 13 deletions(-)

Hello Maxime,

Thanks for this series, that a great step forward. I will have a closer

look later, but as a general remark, should we really target the

core-updates-frozen branch?

Maybe it would make more sense to target the core-updates branch as

there's still some work required to get gtk+ to cross-build.

Mathieu

Mathieu Othacehe schreef op za 28-08-2021 om 15:33 [+0200]:

Going by the patch series title: ‘Support cross-compilation in

glib-or-gtk-build-system’, I don't see why this would have to go in

core-updates instead of core-updates-frozen, because:

(1) no rebuilds, unless I missed something (*)

(2) due to (1), it doesn't introduce bugs

(3) after this patch series, glib-or-gtk-build-system _does_ support

support cross-compilation

(*) In one patch, I unconditionally added a glib:bin input. IIRC, it failed

to build natively previously, but I'd better check again ..

For example, cairo, which uses glib-or-gtk-build-system, now cross-compiles

for aarch64-linux-gnu. True, ideally gtk+ would cross-compile as well,

but cairo being cross-compilable is already useful (e.g., it's a

dependency of harfbuzz which is a dependency of qtbase which is a dependency

of graphical Qt software.) (**)

(**) qtbase (indirectly) depends on perl-file-mimeinfo and 'perl-build-system'

doesn't support cross-compilation, so this argument is probably not really

convincing ... Maybe texlive-bin would be a

better example?(***)

(***) The dependency libungif fails to cross-compile/

Also, hacking on master (or core-updates-frozen) is simpler on than on core-updates,

due to more substitutes, and it is more appealing to me because the cross-compilation

support would be available earlier to most people than if it the patches were based

on core-updates.

As a general principle, I prefer basing patches on 'master'. That wasn't possible

here due to the use of G-exps in glib-or-gtk-build-system, so I based the patches

on 'core-updates-frozen' instead. That goes a bit against

(guix)Submitting Patches though, so I would understand if the patches will have to

be rebased on core-updates:

When we decide to start building the ‘staging’ or ‘core-updates’

branches, they will be forked and renamed with the suffix

‘-frozen’, at which time only bug fixes may be pushed to the frozen

branches.

unless ‘let glib-or-gtk-build-system support cross-compilation’ counts as a bug fix.

To me, this is not all that different from other cross-compilation fixes like

"CC=gcc" --> (string-append "CC=" (cc-for-target)) or moving inputs between

'native-inputs' and 'inputs', but YMMV.

Greetings,

Maxime.

Maxime Devos schreef op wo 25-08-2021 om 19:55 [+0200]:

Be aware that I don't have an aarch64-linux-gnu machine to test this on.

Some binaries of some packages I've run through qemu-aarch64 and I've

run "guix gc --references [...]" on a few store items to check if it

looks reaonable, but that's it.

Greetings,

Maxime.

Hello Maxime,

I made this suggestion mostly because there are quite a few

"TODO(core-updates)" that could be spared by basing this series on

core-updates directly.

Yeah this series is near the limit between a bug fix and a feature

addition. Anyway, let's go for the core-updates-frozen branch if there

are no other objections.

Thanks,

Mathieu

Hey,

Won't this be "cpython-310-" or so when we will update the Python

package definition?

Thanks,

Mathieu

Hey,

Nix is often ahead of Guix regarding cross-compilation support. Did you

have a chance to see how they are dealing with introspection in the

context of cross-compilation?

Thanks,

Mathieu

Hey,

You could maybe extract the files list from the %pkg-config definition

not to duplicate it?

Mathieu

Mathieu Othacehe schreef op ma 30-08-2021 om 15:00 [+0200]:

It appears that, say, atk can be cross-compiled on Nix, but other than that,

Nix doesn't seem to be ahead here. We seem to be in good company here. I found

the following

Nix setting -Dintrospection=false too when cross-compiling:

Nix people finding gobject-introspection is hard to cross-compile

Nix people using an emulator (presumably QEMU?):

Unfortunately, it appears that introspection data is architecture-dependent:

https://archive.md/U8rm2, so just copying over the data from a natively

compiled version won't work (*).

One ‘solution’ is to run g-ir-scanner under QEMU:

https://maxice8.github.io/8-cross-the-gir/. That isn't really cross-compilation

though, but emulated compilation, so it won't work with the Hurd.

Might be good enough for cross-compiling between Linux targets though ...

Here is an alternative idea that I won't attempt in this patch series:

(*) One solution would be to copy it over anyway and substitute things(***) where needed,

and re-generate the typelib data. That seems rather prone to mistakes though,

so maybe we should then hash the adjusted introspection data and compare the hash

to the hash of what we now the introspection data must be (**).

Of course, we can't just guess that, so maybe when compiling atk natively for,

say, x86_64-linux, a message

‘new introspection hashes: ("x86_64-linux-gnu" ("Atk-1.0" "THE-HASH"))’

or

‘stale introspection hash: "Atk-1.0" "THE-OLD-HASH" "THE-NEW-HASH"

followed by:

‘Please add it to #:gir-hashes to help cross-compilers.’

could be printed (maybe even fail the build?).

This hash checking and substitution could be implemented with an additional

build phase for meson-build-system (#:gir would be an additional argument for

meson-build-system).

Something like:

(package

(name "atk")

[...]

(arguments

`(#:gir-hashes

(("x86_64-linux-gnu" ("Atk-1.0" "THE-HASH"))

("powerpc64-linux-gnu"

("Atk-1.0"

;; XXX(core-updates): We forgot to update the hash,

;; and now we're stuck with it unless we want a world-rebuild

;; for powerpc64le-linux (other architectures would be unaffected).

;; Use the right hash when cross-compiling.

,(if (%current-target-system)

"THE-CORRECT-HASH"

"THE-OLD-HASH"))

;; XXX I forgot the triplet, imagine you saw the triplet here instead ...

("i586-hurd" ...)

...)

#:gir-substitutions

;; TODO: need to thing of a good scheme here ...

((replace-value c:identifier "ARCHITECTURE_DEPENDENT"

((? architecture "ppc64le") "x")

((? ...) "y"))

(replace-value c:identifier "KERNEL_DEPENDENT"

((? kernel-os "linux-gnu") "x")

((? kernel-os "gnu") "y"))

;; Only x86_64 has this quirk.

(replace-value c:identifier "QUIRK"

((? architecture "x86_64") "yes)

(otherwise "no"))))

[...])

A limitation of this solution is that someone needs to compile it natively

for the target architecture first (possibly emulated with QEMU).

(**) The .gir includes the name of the shared library, so maybe compute the hash

modulo store references to avoid invalidating the reference hashes every time

an (indirect) dependency is updated.

(***) An ordinary 'substitute*' won't work well, something specific for .gir XML

would be needed. E.g., consider the following (in this case architecture-independent)

snippet from Atk-1.0.gir:

<member name="alert"

value="2"

c:identifier="ATK_ROLE_ALERT"

glib:nick="alert">

<doc xml:space="preserve"

filename="../atk/atkobject.h"

line="39">An object which is an alert to the user. Assistive Technologies typically respond to ATK_ROLE_ALERT by reading the entire onscreen co>

</member>

Sometimes the ‘value="..."’ part is architecture-dependent, and needs to

be adjusted for the target architecture.

Greetings,

Maxime.

Hey,

Thanks for investigating it.

Yes, looks like some projects such as Yocto[1] and Void[2] are also

using this solution. Requiring transparent emulating through QEMU while

cross-compiling will for sure make the process more complex. Restricting

this support to architectures that can be emulated by QEMU is also an

important limitation.

This is a very interesting idea, I'm not familiar with this

introspection mechanism an couldn't judge how hard the substitution

would be. Maybe we could bring this topic upstream, to see what's their

opinion about it.

That's a whole new topic anyway, that shouldn't block the current

patchset we are discussing.

Thanks,

Mathieu

[1]: https://www.openembedded.org/pipermail/openembedded-core/2015-November/113390.html

[2]: https://voidlinux.org/news/2019/02/cross-gobject.html

Mathieu Othacehe schreef op ma 30-08-2021 om 15:13 [+0200]:

Done in local checkout. Also, the package inherits from %pkg-config instead

of pkg-config now (doesn't change the derivations but seems more proper).

Let me now when I should send a v2 (I'll look into the cpython-39 issue first).

Greetings,

Maxime

Mathieu Othacehe schreef op ma 30-08-2021 om 14:58 [+0200]:

Indeed. I attached a patch independent of python's version number,

to be squashed in the v2 patch series.

Greetings,

Maxime

Mathieu Othacehe schreef op di 31-08-2021 om 11:49 [+0200]:

FWIW, IIUC, qemu-binfmt-style _transparent_ emulation is not required. "meson" can

be told where to find the QEMU binary, with an option in the ‘cross file’.

Greetings,

Maxime.

Hello Maxime,

Nice, thanks. The series doesn't apply anymore, would you have a branch

somewhere that I could use?

Mathieu

Mathieu Othacehe schreef op zo 19-09-2021 om 15:43 [+0000]:

The merge conflicts were trivial (in the copyright lines).

Anyway, you could find a rebased branch at https://notabug.org/maximed/guix-gnunet/src/gtk-cross-fixes1

(untested) (commit: 3936dfc4d1c29bb5311caa3597ca272c8326f0c3).

Greetings,

Maxime.

Hello,

I pushed the series, with a minor modification to add a reference to our

discussion on introspection data cross-compilation.

It would be nice to remove the associated TODOs on core-updates while

this is still fresh.

Thanks for your work on that topic,

Mathieu

Hi,

Maxime Devos <maximedevos@telenet.be> skribis:

Isn’t that a typo, “PKG_CONFIG_FOR_BUILD” instead of

“PKG_CONFIG_PATH_FOR_BUILD” or (string-append name "_FOR_BUILD")?

The original spec above looked more appropriate to me.

Thanks,

Ludo’.

Ludovic Courtès schreef op wo 22-09-2021 om 21:10 [+0200]:

Yes, it's a typo. The attached patch should fix it (now running

./pre-inst-env guix build --target=... ...). I wonder how I could have

missed this.

Greetings,

Maxime.

Maxime Devos schreef op do 23-09-2021 om 12:38 [+0200]:

No pkg-config related build failures when doing

‘./pre-inst-env guix build --target=... gtk+’ so far.