[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Tobias Geerinckx-Rice]

* gnu/packages/networking.scm (whois)[source]: Remove mkpasswd sources in a
snippet.
[native-inputs]: Add gettext.
[arguments]: Add ‘update-po’ phase.
[description]: Remove historical not about bundled mkpasswd.
---

Guix,

A quick TODO item glanced in a recent commit notification, done.

While everything could be done in a single phase, I prefer ripping such
cruft out early. Manually building the modified sources works fine, at
the cost of slightly bloated .po files with some unused strings. It's
unlikely someone will want to unpack this to build mkpasswd...

Then again, it's not a licence issue so I don't really care that much.

Kind regards,

T G-R

 gnu/packages/networking.scm | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)

diff --git a/gnu/packages/networking.scm b/gnu/packages/networking.scm
index 94621052d..9a8993be2 100644
--- a/gnu/packages/networking.scm
+++ b/gnu/packages/networking.scm
@@ -4,7 +4,7 @@
 ;;; Copyright © 2015 Mark H Weaver <mhw@netris.org>
 ;;; Copyright © 2015, 2016, 2017 Stefan Reichör <stefan@xsteve.at>
 ;;; Copyright © 2016 Raimon Grau <raimonster@gmail.com>
-;;; Copyright © 2016, 2017 Tobias Geerinckx-Rice <me@tobias.gr>
+;;; Copyright © 2016, 2017, 2018 Tobias Geerinckx-Rice <me@tobias.gr>
 ;;; Copyright © 2016 John Darrington <jmd@gnu.org>
 ;;; Copyright © 2016, 2017 Nicolas Goaziou <mail@nicolasgoaziou.fr>
 ;;; Copyright © 2016 Eric Bavier <bavier@member.fsf.org>
@@ -458,15 +458,31 @@ and up to 1 Mbit/s downstream.")
                            name "_" version ".tar.xz"))
        (sha256
         (base32
-         "02f00vpgrdb77w7lskl9jfm2akpy21ws9cjazs13gash2xksnj38"))))
+         "02f00vpgrdb77w7lskl9jfm2akpy21ws9cjazs13gash2xksnj38"))
+       (modules '((guix build utils)))
+       (snippet
+        '(begin
+           ;; Remove unused ‘mkpasswd’ bundled for historical reasons.
+           (for-each delete-file
+                     (find-files "." "^mkpasswd\\."))
+           (substitute* "Makefile"
+             ((" (install-|)mkpasswd ") " "))
+           (substitute* "po/Makefile"
+             (("\\.\\./mkpasswd\\.c") ""))
+           #t))))
     (build-system gnu-build-system)
-    ;; TODO: unbundle mkpasswd binary + its po files.
+    (native-inputs
+     `(("gettext" ,gettext)))           ; for update-po phase
     (arguments
      `(#:tests? #f ; Does not exist
        #:make-flags (list "CC=gcc"
                           (string-append "prefix=" (assoc-ref %outputs "out")))
        #:phases
        (modify-phases %standard-phases
+         (add-after 'unpack 'update-po
+           ;; Remove harmless but useless ‘mkpasswd’ remnants from .po files.
+           (lambda _
+             (invoke "make" "-C" "po" "update-po")))
          (delete 'configure) ; No configure
          (add-before 'build 'setenv
            (lambda _
@@ -481,9 +497,7 @@ and up to 1 Mbit/s downstream.")
        ("pkg-config" ,pkg-config)))
     (synopsis "Improved whois client")
     (description "This whois client is intelligent and can
-automatically select the appropriate whois server for most queries.
-Because of historical reasons this also includes a tool called mkpasswd
-which can be used to encrypt a password with @code{crypt(3)}.")
+automatically select the appropriate whois server for most queries.")
     (home-page "https://github.com/rfc1036/whois")
     (license license:gpl2+)))
 
-- 
2.15.0

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Tobias Geerinckx-Rice]

Tobias Geerinckx-Rice wrote on 09/01/18 at 02:01:
> * gnu/packages/networking.scm (whois)[source]: Remove mkpasswd sources in a
> snippet.
> [native-inputs]: Add gettext.
> [arguments]: Add ‘update-po’ phase.
> [description]: Remove historical not about bundled mkpasswd.
                                   ^^^
                                   Sigh. Noted.

Kind regards,

T G-R

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[ng0]

[-- Attachment #1: Type: text/plain, Size: 4163 bytes --]

Tobias Geerinckx-Rice transcribed 3.4K bytes:
> * gnu/packages/networking.scm (whois)[source]: Remove mkpasswd sources in a
> snippet.
> [native-inputs]: Add gettext.
> [arguments]: Add ‘update-po’ phase.
> [description]: Remove historical not about bundled mkpasswd.
> ---
> 
> Guix,
> 
> A quick TODO item glanced in a recent commit notification, done.
> 
> While everything could be done in a single phase, I prefer ripping such
> cruft out early. Manually building the modified sources works fine, at
> the cost of slightly bloated .po files with some unused strings. It's
> unlikely someone will want to unpack this to build mkpasswd...

or you could move the mkpasswd to its own output. I think you want to
look into the old discussions around this. I don't remember why this
wasn't chosen, all I know is that upstream made it very clear to me
that they won't separate mkpasswd.

> 
> Then again, it's not a licence issue so I don't really care that much.
> 
> Kind regards,
> 
> T G-R
> 
>  gnu/packages/networking.scm | 26 ++++++++++++++++++++------
>  1 file changed, 20 insertions(+), 6 deletions(-)
> 
> diff --git a/gnu/packages/networking.scm b/gnu/packages/networking.scm
> index 94621052d..9a8993be2 100644
> --- a/gnu/packages/networking.scm
> +++ b/gnu/packages/networking.scm
> @@ -4,7 +4,7 @@
>  ;;; Copyright © 2015 Mark H Weaver <mhw@netris.org>
>  ;;; Copyright © 2015, 2016, 2017 Stefan Reichör <stefan@xsteve.at>
>  ;;; Copyright © 2016 Raimon Grau <raimonster@gmail.com>
> -;;; Copyright © 2016, 2017 Tobias Geerinckx-Rice <me@tobias.gr>
> +;;; Copyright © 2016, 2017, 2018 Tobias Geerinckx-Rice <me@tobias.gr>
>  ;;; Copyright © 2016 John Darrington <jmd@gnu.org>
>  ;;; Copyright © 2016, 2017 Nicolas Goaziou <mail@nicolasgoaziou.fr>
>  ;;; Copyright © 2016 Eric Bavier <bavier@member.fsf.org>
> @@ -458,15 +458,31 @@ and up to 1 Mbit/s downstream.")
>                             name "_" version ".tar.xz"))
>         (sha256
>          (base32
> -         "02f00vpgrdb77w7lskl9jfm2akpy21ws9cjazs13gash2xksnj38"))))
> +         "02f00vpgrdb77w7lskl9jfm2akpy21ws9cjazs13gash2xksnj38"))
> +       (modules '((guix build utils)))
> +       (snippet
> +        '(begin
> +           ;; Remove unused ‘mkpasswd’ bundled for historical reasons.
> +           (for-each delete-file
> +                     (find-files "." "^mkpasswd\\."))
> +           (substitute* "Makefile"
> +             ((" (install-|)mkpasswd ") " "))
> +           (substitute* "po/Makefile"
> +             (("\\.\\./mkpasswd\\.c") ""))
> +           #t))))
>      (build-system gnu-build-system)
> -    ;; TODO: unbundle mkpasswd binary + its po files.
> +    (native-inputs
> +     `(("gettext" ,gettext)))           ; for update-po phase
>      (arguments
>       `(#:tests? #f ; Does not exist
>         #:make-flags (list "CC=gcc"
>                            (string-append "prefix=" (assoc-ref %outputs "out")))
>         #:phases
>         (modify-phases %standard-phases
> +         (add-after 'unpack 'update-po
> +           ;; Remove harmless but useless ‘mkpasswd’ remnants from .po files.
> +           (lambda _
> +             (invoke "make" "-C" "po" "update-po")))
>           (delete 'configure) ; No configure
>           (add-before 'build 'setenv
>             (lambda _
> @@ -481,9 +497,7 @@ and up to 1 Mbit/s downstream.")
>         ("pkg-config" ,pkg-config)))
>      (synopsis "Improved whois client")
>      (description "This whois client is intelligent and can
> -automatically select the appropriate whois server for most queries.
> -Because of historical reasons this also includes a tool called mkpasswd
> -which can be used to encrypt a password with @code{crypt(3)}.")
> +automatically select the appropriate whois server for most queries.")
>      (home-page "https://github.com/rfc1036/whois")
>      (license license:gpl2+)))
>  
> -- 
> 2.15.0
> 
> 
> 
> 
> 

-- 
GnuPG: A88C8ADD129828D7EAC02E52E22F9BBFEE348588
GnuPG: https://c.n0.is/ng0_pubkeys/tree/keys
  WWW: https://n0.is/a/  ::  https://ea.n0.is

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

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Ludovic Courtès]

Tobias Geerinckx-Rice <me@tobias.gr> skribis:

> * gnu/packages/networking.scm (whois)[source]: Remove mkpasswd sources in a
> snippet.
> [native-inputs]: Add gettext.
> [arguments]: Add ‘update-po’ phase.
> [description]: Remove historical not about bundled mkpasswd.

LGTM, thanks!

Ludo'.

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Tobias Geerinckx-Rice]

[-- Attachment #1.1: Type: text/plain, Size: 1393 bytes --]

ng0,

ng0 wrote on 09/01/18 at 07:51:
> or you could move the mkpasswd to its own output.

That's not (a step towards) unbundling, though, is it?

> I think you want to look into the old discussions around this. I
> don't remember why this wasn't chosen, all I know is that upstream
> made it very clear to me that they won't separate mkpasswd.

OK, I found this thread[0] on guix-devel. I'd only checked guix-patches.

I can somewhat understand your use of the word ‘bundle’ in the TODO
comment — or at least where it came from — but it's pretty misleading in
that context. As you rightly noted:

  Ludo'> Why to you call that “unbundling”?
  ng0>   Wrong choice of words.
  [...]
  ng0>   It's annoying that two unrelated applications are thrown
  ng0>   into one distribution/package.

It would have been nice to adjust the comment at that point, or mention
that this ‘mkpasswd’ is unrelated to the better-known Expect ‘mkpasswd’.

(Hark, this one doesn't even mk passwds! For maximum irony, it was
 renamed from ‘cryptpw’, which perfectly described its purpose...)

Since this is its own thing, for better or worse, I agree with the
others in that thread that it should remain part of this package/output
until upstream repents.

Kind regards,

T G-R

[0]: https://lists.gnu.org/archive/html/guix-devel/2017-01/msg00504.html


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 248 bytes --]

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Tobias Geerinckx-Rice]

Ludo',

Ludovic Courtès wrote on 09/01/18 at 14:36:
> LGTM, thanks!

Thanks. However, I'm now convinced that this program is confusingly
named but belongs where it is, and we should simply remove the TODO
comment. It's not ‘bundled’ in the sense of the word we use.

If no-one objects I'll do just that & close this bug.

Kind regards,

T G-R

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 650 bytes --]

Tobias Geerinckx-Rice <me@tobias.gr> writes:

> Ludo',
>
> Ludovic Courtès wrote on 09/01/18 at 14:36:
>> LGTM, thanks!
>
> Thanks. However, I'm now convinced that this program is confusingly
> named but belongs where it is, and we should simply remove the TODO
> comment. It's not ‘bundled’ in the sense of the word we use.
>
> If no-one objects I'll do just that & close this bug.

LGTM!

I was just about to write that I actually *use* this software, and
suggest doing the same thing.  It's a convenient tool for generating a
password string suitable for /etc/shadow.  And it's tiny, so separating
it out seems "overkill".

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

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[ng0]

[-- Attachment #1: Type: text/plain, Size: 1258 bytes --]

Marius Bakke transcribed 1.3K bytes:
> Tobias Geerinckx-Rice <me@tobias.gr> writes:
> 
> > Ludo',
> >
> > Ludovic Courtès wrote on 09/01/18 at 14:36:
> >> LGTM, thanks!
> >
> > Thanks. However, I'm now convinced that this program is confusingly
> > named but belongs where it is, and we should simply remove the TODO
> > comment. It's not ‘bundled’ in the sense of the word we use.
> >
> > If no-one objects I'll do just that & close this bug.
> 
> LGTM!
> 
> I was just about to write that I actually *use* this software, and
> suggest doing the same thing.  It's a convenient tool for generating a
> password string suitable for /etc/shadow.  And it's tiny, so separating
> it out seems "overkill".

Which is exactly what we concluded in the old thread about this, which
I suggested to read. I just should've removed the TODO note a long time
ago.
I still find mkpasswd feels like an historical accident bundling in favor
of some OS that relies on it (iirc the old thread or authors replies),
but doesn't really matter. Bikeshedding on the highest level of bikesheds
I guess.
-- 
GnuPG: A88C8ADD129828D7EAC02E52E22F9BBFEE348588
GnuPG: https://c.n0.is/ng0_pubkeys/tree/keys
  WWW: https://n0.is/a/  ::  https://ea.n0.is

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

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[ng0]

[-- Attachment #1: Type: text/plain, Size: 1974 bytes --]

Tobias Geerinckx-Rice transcribed 2.6K bytes:
> ng0,
> 
> ng0 wrote on 09/01/18 at 07:51:
> > or you could move the mkpasswd to its own output.
> 
> That's not (a step towards) unbundling, though, is it?
> 
> > I think you want to look into the old discussions around this. I
> > don't remember why this wasn't chosen, all I know is that upstream
> > made it very clear to me that they won't separate mkpasswd.
> 
> OK, I found this thread[0] on guix-devel. I'd only checked guix-patches.
> 
> I can somewhat understand your use of the word ‘bundle’ in the TODO
> comment — or at least where it came from — but it's pretty misleading in
> that context. As you rightly noted:
> 
>   Ludo'> Why to you call that “unbundling”?
>   ng0>   Wrong choice of words.
>   [...]
>   ng0>   It's annoying that two unrelated applications are thrown
>   ng0>   into one distribution/package.
> 
> It would have been nice to adjust the comment at that point, or mention
> that this ‘mkpasswd’ is unrelated to the better-known Expect ‘mkpasswd’.
> 
> (Hark, this one doesn't even mk passwds! For maximum irony, it was
>  renamed from ‘cryptpw’, which perfectly described its purpose...)
> 
> Since this is its own thing, for better or worse, I agree with the
> others in that thread that it should remain part of this package/output
> until upstream repents.
> 
> Kind regards,
> 
> T G-R
> 
> [0]: https://lists.gnu.org/archive/html/guix-devel/2017-01/msg00504.html
> 

I had whois installed for some time, but consider this (or test it):

What if someone under account root install 'whois' into their profile.
Wouldn't they have 2 mkpasswd now in their path of exectuables and
encounter a collision?
Should we rename it for the sake of namespace collisions reduction?

-- 
GnuPG: A88C8ADD129828D7EAC02E52E22F9BBFEE348588
GnuPG: https://c.n0.is/ng0_pubkeys/tree/keys
  WWW: https://n0.is/a/  ::  https://ea.n0.is

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

bug#30037: [PATCH] gnu: whois: Remove bundled mkpasswd.

[Tobias Geerinckx-Rice]

[-- Attachment #1.1: Type: text/plain, Size: 664 bytes --]

ng0 wrote on 10/01/18 at 15:46:
> Which is exactly what we concluded in the old thread about this,
> which I suggested to read. I just should've removed the TODO note a
> long time ago.

As I wrote in my previous message addressing that exact thread, that
would have been nice. No harm done beyond some wasted time.

> I still find mkpasswd feels like an historical accident bundling in
> favor of some OS that relies on it (iirc the old thread or authors
> replies), but doesn't really matter. Bikeshedding on the highest
> level of bikesheds I guess.

Yup :-) I've removed the comment in master and am closing the shed.

Kind regards,

T G-R


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 248 bytes --]

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[Tobias Geerinckx-Rice]

[-- Attachment #1.1: Type: text/plain, Size: 789 bytes --]

ng0,

ng0 wrote on 10/01/18 at 15:55:
> What if someone under account root install 'whois' into their 
> profile.

[together with expect, source of The Other Mkpasswd.]

> Wouldn't they have 2 mkpasswd now in their path of exectuables and 
> encounter a collision?

Presumably.

> Should we rename it for the sake of namespace collisions reduction?

I don't think this case warrants manual intervention.

Whois' mkpasswd encrypt()s a given passwd and prints the result.
Expect's generates a random string that satisfies some criteria.
Neither are what I'd call day-to-day system administration tools.

I'd rather give users deterministic control over any collision, but
now who's bikeshedding (with no intention to write the code)... :-)

Kind regards,

T G-R


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 248 bytes --]

[bug#30037] [PATCH] gnu: whois: Remove bundled mkpasswd.

[ng0]

[-- Attachment #1: Type: text/plain, Size: 913 bytes --]

Tobias Geerinckx-Rice transcribed 2.0K bytes:
> ng0 wrote on 10/01/18 at 15:46:
> > Which is exactly what we concluded in the old thread about this,
> > which I suggested to read. I just should've removed the TODO note a
> > long time ago.
> 
> As I wrote in my previous message addressing that exact thread, that
> would have been nice. No harm done beyond some wasted time.
> 
> > I still find mkpasswd feels like an historical accident bundling in
> > favor of some OS that relies on it (iirc the old thread or authors
> > replies), but doesn't really matter. Bikeshedding on the highest
> > level of bikesheds I guess.
> 
> Yup :-) I've removed the comment in master and am closing the shed.
> 
> Kind regards,
> 
> T G-R
> 


Okay, ty!

-- 
GnuPG: A88C8ADD129828D7EAC02E52E22F9BBFEE348588
GnuPG: https://c.n0.is/ng0_pubkeys/tree/keys
  WWW: https://n0.is/a/  ::  https://ea.n0.is

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