Document GUIX_LOAD_PATH?

Document GUIX_LOAD_PATH?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 214 bytes --]

Hi!

Today I learned about the existence of GUIX_LOAD_PATH.  It does not seem
to be documented in the manual.  Is it anywhere in the doc?  Shall we
document it?

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: Document GUIX_LOAD_PATH?

[zimoun]

Hi Pierre,

On Wed, 18 Dec 2019 at 12:28, Pierre Neidhardt <mail@ambrevar.xyz> wrote:

> Today I learned about the existence of GUIX_LOAD_PATH.  It does not seem
> to be documented in the manual.  Is it anywhere in the doc?  Shall we
> document it?

It can be documented in devel pages., maybe Contributing.

But IMHO, the normal use should be via the --load-path option and the
GUIX_LOAD_PATH should be less and less used; stay here for historical
reason and/or backward compatibility and/or some devel use-case.

Well, for discoverability, it appears to me easier to list an option
with "guix <cmd> --help" than list environment variables.


All the best,
simon

Re: Document GUIX_LOAD_PATH?

[Clément Lassieur]

zimoun <zimon.toutoune@gmail.com> writes:

> Hi Pierre,
>
> On Wed, 18 Dec 2019 at 12:28, Pierre Neidhardt <mail@ambrevar.xyz> wrote:
>
>> Today I learned about the existence of GUIX_LOAD_PATH.  It does not seem
>> to be documented in the manual.  Is it anywhere in the doc?  Shall we
>> document it?
>
> It can be documented in devel pages., maybe Contributing.
>
> But IMHO, the normal use should be via the --load-path option and the
> GUIX_LOAD_PATH should be less and less used; stay here for historical
> reason and/or backward compatibility and/or some devel use-case.
>
> Well, for discoverability, it appears to me easier to list an option
> with "guix <cmd> --help" than list environment variables.

Hi Simon and Pierre,

It was me who talked to you about it, and...  It's not GUIX_LOAD_PATH,
it's GUIX_PACKAGE_PATH, sorry!!

(And it's documented :))

See you soon!
Clément

Re: Document GUIX_LOAD_PATH?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 83 bytes --]

Hahaha, sorry for the noise then!

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: Document GUIX_LOAD_PATH?

[Ludovic Courtès]

Also, GUIX_PACKAGE_PATH is not “recommended”—channels FTW!

Ludo’.

Re: Document GUIX_LOAD_PATH?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 512 bytes --]

The reason this came up was because the --load-path command line
argument was missing from some command.  Then Clément argued that it was
not necessary anyways, we could set GUIX_PACKAGE_PATH (I didn't know
that it was effectively the same as --load-path, but for all commands).

So the question is: shall we add --load-path to all commands instead?
(If not already the case.)

Side question: Can we really do everything without using GUIX_PACKAGE_PATH?

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: Document GUIX_LOAD_PATH?

[zimoun]

On Thu, 19 Dec 2019 at 18:04, Pierre Neidhardt <mail@ambrevar.xyz> wrote:

> So the question is: shall we add --load-path to all commands instead?
> (If not already the case.)

Only commands that need it. :-)

Do you have in mind commands where the option --load-path is still missing?
Maybe "guix graph"?
Maybe "guix edit"?

Re: Document GUIX_LOAD_PATH?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 434 bytes --]

zimoun <zimon.toutoune@gmail.com> writes:

> Do you have in mind commands where the option --load-path is still missing?
> Maybe "guix graph"?

Yes.

> Maybe "guix edit"?

Yes.

Also:

- guix size (this one has annoyed me for a long time!)

- Maybe guix import?  What if a channel / local repo specifies a new
  importer?

- guix refresh

I think that would be it.

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: Document GUIX_LOAD_PATH?

[zimoun]

Hi,

On Thu, 19 Dec 2019 at 18:04, Pierre Neidhardt <mail@ambrevar.xyz> wrote:

> So the question is: shall we add --load-path to all commands instead?
> (If not already the case.)

Please comment patches #38678.

https://debbugs.gnu.org/cgi/bugreport.cgi?bug=38678


All the best,
simon

Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Danny Milosavljevic]

[-- Attachment #1: Type: text/plain, Size: 932 bytes --]

Hi Pierre,
Hi zimoun,

On Thu, 19 Dec 2019 18:14:53 +0100
Pierre Neidhardt <mail@ambrevar.xyz> wrote:

> zimoun <zimon.toutoune@gmail.com> writes:
> 
> > Do you have in mind commands where the option --load-path is still missing?
> > Maybe "guix graph"?  
> 
> Yes.
> 
> > Maybe "guix edit"?  
> 
> Yes.

... why?  There's the environment variable already--what does it give us
to have yet another way to specify the value?

Shells allow you to do

  ENVVAR1=value command1

anyway.

zimoun wrote:

>But IMHO, the normal use should be via the --load-path option and the
>GUIX_LOAD_PATH should be less and less used; stay here for historical
>reason and/or backward compatibility and/or some devel use-case.

Why?  *PATH environment variables have a long tradition in UNIX and are not
going to vanish (or mean something else any time) soon.  What is the use
case this change is supposed to enable?

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 393 bytes --]

Hi Danny!

The environment variable is deprecated if I understand properly.
Either way, we need to be consistent.  Some subcommands support
--load-path while others don't, for no apparent reason.  I believe it's
a good idea to minimize the element of surprise and conversely maximize
familiarity with the command line options.

Cheers!

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[zimoun]

Hi Danny,

Thank you for your insights.

On Thu, 26 Dec 2019 at 21:31, Danny Milosavljevic
<dannym@scratchpost.org> wrote:

> ... why?  There's the environment variable already--what does it give us
> to have yet another way to specify the value?

From my opinion, the answers to the why is:
 - because it is easier to explain to newcomers; only a new option and
not a new concept,
 - because it adds consistency with other commands, as Pierre said,
 - because it is easier to discover via "guix <cmd> -h"; even if it
should also be easy to add environment variables documentation via
"guix <cmd> -h".


> zimoun wrote:
>
> >But IMHO, the normal use should be via the --load-path option and the
> >GUIX_LOAD_PATH should be less and less used; stay here for historical
> >reason and/or backward compatibility and/or some devel use-case.
>
> Why?  *PATH environment variables have a long tradition in UNIX and are not
> going to vanish (or mean something else any time) soon.  What is the use
> case this change is supposed to enable?

Why? I do not know, maybe gray beard vs young mustache. ;-)
The option "--load-path" is more a flavor of taste than an use case
uncovered by GUIX_PACKAGE_PATH.
And GUIX_PACKAGE_PATH is not going to vanish.


Hope that helps.

All the best,
simon

Re: Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 1070 bytes --]

I've merged your patch 38678.
Only one nit with guix refresh:

--8<---------------cut here---------------start------------->8---
+        ;; The short option -L is already used by --list-updaters, therefore
+        ;; it needs to be removed from %standard-build-options.
+        (let ((%load-path-option (find (lambda (option)
+                                         (member "load-path"
+                                                 (option-names option)))
+                                       %standard-build-options)))
+          (option
+           (filter (lambda (name) (not (equal? #\L name)))
+                   (option-names %load-path-option))
+           (option-required-arg? %load-path-option)
+           (option-optional-arg? %load-path-option)
+           (option-processor     %load-path-option)))
+
--8<---------------cut here---------------end--------------->8---

I would name the let-bound variable without a percent sign '%' by
convention.
Should I fix this for you?

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[zimoun]

[-- Attachment #1: Type: text/plain, Size: 203 bytes --]

On Thu, 16 Jan 2020 at 15:30, Pierre Neidhardt <mail@ambrevar.xyz> wrote:

> I've merged your patch 38678.

Cool! Thank you.


> Should I fix this for you?

Yes, for sure.
Patch attached.

Cheers,
simon

[-- Attachment #2: v5-0003-refresh-Add-load-path-option.patch --]
[-- Type: text/x-patch, Size: 3577 bytes --]

From 7ec82e221fdb42b6776ae4c41708d9ec361b4ebf Mon Sep 17 00:00:00 2001
From: zimoun <zimon.toutoune@gmail.com>
Date: Wed, 8 Jan 2020 18:53:01 +0100
Subject: [PATCH v5 3/6] refresh: Add '--load-path' option.

* guix/scripts/refresh.scm (%option): Add '--load-path' option.
* doc/guix.texi: Document it.
---
 doc/guix.texi            |  9 ++++++++-
 guix/scripts/refresh.scm | 18 ++++++++++++++++++
 2 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/doc/guix.texi b/doc/guix.texi
index 3d26facd2e..bee4c3b2c9 100644
--- a/doc/guix.texi
+++ b/doc/guix.texi
@@ -69,7 +69,7 @@ Copyright @copyright{} 2019 Jakob L. Kreuze@*
 Copyright @copyright{} 2019 Kyle Andrews@*
 Copyright @copyright{} 2019 Alex Griffin@*
 Copyright @copyright{} 2019 Guillaume Le Vaillant@*
-Copyright @copyright{} 2019 Simon Tournier@*
+Copyright @copyright{} 2019, 2020 Simon Tournier@*
 
 Permission is granted to copy, distribute and/or modify this document
 under the terms of the GNU Free Documentation License, Version 1.3 or
@@ -9552,6 +9552,13 @@ the user whether to download it or not.  This is the default behavior.
 @item --key-server=@var{host}
 Use @var{host} as the OpenPGP key server when importing a public key.
 
+@item --load-path=@var{directory}
+Add @var{directory} to the front of the package module search path
+(@pxref{Package Modules}).
+
+This allows users to define their own packages and make them visible to
+the command-line tools.
+
 @end table
 
 The @code{github} updater uses the
diff --git a/guix/scripts/refresh.scm b/guix/scripts/refresh.scm
index daf6fcf947..efada1df5a 100644
--- a/guix/scripts/refresh.scm
+++ b/guix/scripts/refresh.scm
@@ -7,6 +7,7 @@
 ;;; Copyright © 2017 Mathieu Othacehe <m.othacehe@gmail.com>
 ;;; Copyright © 2018 Efraim Flashner <efraim@flashner.co.il>
 ;;; Copyright © 2019 Ricardo Wurmus <rekado@elephly.net>
+;;; Copyright © 2020 Simon Tournier <zimon.toutoune@gmail.com>
 ;;;
 ;;; This file is part of GNU Guix.
 ;;;
@@ -27,6 +28,7 @@
   #:use-module (guix ui)
   #:use-module (gcrypt hash)
   #:use-module (guix scripts)
+  #:use-module ((guix scripts build) #:select (%standard-build-options))
   #:use-module (guix store)
   #:use-module (guix utils)
   #:use-module (guix packages)
@@ -116,6 +118,19 @@
                      (leave (G_ "unsupported policy: ~a~%")
                             arg)))))
 
+        ;; The short option -L is already used by --list-updaters, therefore
+        ;; it needs to be removed from %standard-build-options.
+        (let ((load-path-option (find (lambda (option)
+                                         (member "load-path"
+                                                 (option-names option)))
+                                       %standard-build-options)))
+          (option
+           (filter (lambda (name) (not (equal? #\L name)))
+                   (option-names load-path-option))
+           (option-required-arg? load-path-option)
+           (option-optional-arg? load-path-option)
+           (option-processor     load-path-option)))
+
         (option '(#\h "help") #f #f
                 (lambda args
                   (show-help)
@@ -165,6 +180,9 @@ specified with `--select'.\n"))
                          'always', 'never', and 'interactive', which is also
                          used when 'key-download' is not specified"))
   (newline)
+  (display (G_ "
+      --load-path=DIR    prepend DIR to the package module search path"))
+  (newline)
   (display (G_ "
   -h, --help             display this help and exit"))
   (display (G_ "
-- 
2.23.0

Re: Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 119 bytes --]

I have already merged, can you rebase your changes against master? :)

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[zimoun]

On Fri, 17 Jan 2020 at 16:56, Pierre Neidhardt <mail@ambrevar.xyz> wrote:
>
> I have already merged, can you rebase your changes against master? :)

By "merged", you mean pushed. ;-)

It is not a rebase, but another commit, right?
What will be the commit message?

--8<---------------cut here---------------start------------->8---
refresh: Fix internal variable name.

* guix/scripts/refresh.scm (%option): Fix internal variable name.
--8<---------------cut here---------------stop-------------->8---

It is ok?

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 61 bytes --]

Sure thing!

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[zimoun]

[-- Attachment #1: Type: text/plain, Size: 539 bytes --]

What mess did you do Pierre? :-)

Why are you the author of ee9a735bc8 [1]?
Why there are lines modified in guix.texi 21f4fbdd84 [2] which are not
in the original patch [3]?

[1] https://git.savannah.gnu.org/cgit/guix.git/commit/?id=ee9a735bc8f544cf8eedc6c6a7e4ed2962663013
[2] https://git.savannah.gnu.org/cgit/guix.git/commit/?id=21f4fbdd8453e489fb89825c4226a0a0bda2bc17
[3] https://debbugs.gnu.org/cgi/bugreport.cgi?bug=38678#65

Or why I have missed?

Attached the change fix your early push of the 'refresh' patch. :-)

Cheers,
simon

[-- Attachment #2: 0001-refresh-Fix-internal-variable-name.patch --]
[-- Type: text/x-patch, Size: 1602 bytes --]

From 8dfcf205f023b609117f3da9007e830df406357b Mon Sep 17 00:00:00 2001
From: zimoun <zimon.toutoune@gmail.com>
Date: Fri, 17 Jan 2020 18:30:00 +0100
Subject: [PATCH 1/1] refresh: Fix internal variable name.

* guix/scripts/refresh.scm (%option): Fix internal variable name.
---
 guix/scripts/refresh.scm | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/guix/scripts/refresh.scm b/guix/scripts/refresh.scm
index bc8e906054..efada1df5a 100644
--- a/guix/scripts/refresh.scm
+++ b/guix/scripts/refresh.scm
@@ -120,16 +120,16 @@
 
         ;; The short option -L is already used by --list-updaters, therefore
         ;; it needs to be removed from %standard-build-options.
-        (let ((%load-path-option (find (lambda (option)
+        (let ((load-path-option (find (lambda (option)
                                          (member "load-path"
                                                  (option-names option)))
                                        %standard-build-options)))
           (option
            (filter (lambda (name) (not (equal? #\L name)))
-                   (option-names %load-path-option))
-           (option-required-arg? %load-path-option)
-           (option-optional-arg? %load-path-option)
-           (option-processor     %load-path-option)))
+                   (option-names load-path-option))
+           (option-required-arg? load-path-option)
+           (option-optional-arg? load-path-option)
+           (option-processor     load-path-option)))
 
         (option '(#\h "help") #f #f
                 (lambda args
-- 
2.23.0

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 445 bytes --]

zimoun <zimon.toutoune@gmail.com> writes:

> What mess did you do Pierre? :-)
>
> Why are you the author of ee9a735bc8 [1]?
> Why there are lines modified in guix.texi 21f4fbdd84 [2] which are not
> in the original patch [3]?

I just fixed the conflict of the copyright line because it got updated
in the meantime.  That's it :)

I'm the committer of the patch, not the author though.

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 518 bytes --]

zimoun <zimon.toutoune@gmail.com> writes:

> What mess did you do Pierre? :-)
>
> Why are you the author of ee9a735bc8 [1]?

Good question :(  My apologies, it seems that when I resolve the
conflict your authorship was lost in the process.  Sorry about that.

> Why there are lines modified in guix.texi 21f4fbdd84 [2] which are not
> in the original patch [3]?

Hmmm... Looks like Emacs' ws-buttler had a hickup here.  I don't know
why, sorry about that.

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 91 bytes --]

And I've merged your last patch.  Thanks!

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[zimoun]

On Sat, 18 Jan 2020 at 13:01, Pierre Neidhardt <mail@ambrevar.xyz> wrote:

> > Why are you the author of ee9a735bc8 [1]?
>
> Good question :(  My apologies, it seems that when I resolve the
> conflict your authorship was lost in the process.  Sorry about that.

You owe me a beer at Guix Days 1. ;-)


> > Why there are lines modified in guix.texi 21f4fbdd84 [2] which are not
> > in the original patch [3]?
>
> Hmmm... Looks like Emacs' ws-buttler had a hickup here.  I don't know
> why, sorry about that.

Well, you owe me a beer at Guix Days 2. ;-)


Yes, extra spaces had been introduced by these commits:
21531add320
83db0205060
The good point is now, it is fixed. ;-)


What I do not understand is: why 'ws-buttler' had a hiccup? Did you
modify my patch? I mean ws-buttler generally works by hooking
(before-save-hook), therefore to have a hiccup, 'ws-buttler' needed a
modification then a save, right? Why?


Thank you for reviewing and pushing.
I will remind that you owe me 2 belgian beers. ;-)


Cheers,
simon

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[Pierre Neidhardt]

[-- Attachment #1: Type: text/plain, Size: 528 bytes --]

zimoun <zimon.toutoune@gmail.com> writes:

> What I do not understand is: why 'ws-buttler' had a hiccup? Did you
> modify my patch? I mean ws-buttler generally works by hooking
> (before-save-hook), therefore to have a hiccup, 'ws-buttler' needed a
> modification then a save, right? Why?

I think this happened when I resolved the conflict (with Ediff).

> Thank you for reviewing and pushing.
> I will remind that you owe me 2 belgian beers. ;-)

Stella Artois? :D

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

[bug#38678] Command line option in addition to GUIX_PACKAGE_PATH - Reason?

[zimoun]

On Mon, 20 Jan 2020 at 19:12, Pierre Neidhardt <mail@ambrevar.xyz> wrote:
>
> zimoun <zimon.toutoune@gmail.com> writes:
>
> > What I do not understand is: why 'ws-buttler' had a hiccup? Did you
> > modify my patch? I mean ws-buttler generally works by hooking
> > (before-save-hook), therefore to have a hiccup, 'ws-buttler' needed a
> > modification then a save, right? Why?
>
> I think this happened when I resolved the conflict (with Ediff).

You have not resolved a conflict for this one. It was for the previous
one. That's why I do not understand.

But it does not matter. Only the beers you owe me matter ;-)

Cheer,
simon