bug#56432: run-dig vs dig -- do we need both?

bug#56432: run-dig vs dig -- do we need both?

[Stefan Kangas]

Severity: wishlist

Do we really need both the `dig' and the `run-dig' commands?

Should one of them be obsoleted?  If we want to be able to specify both
query-type and server, we could consider using double and triple
prefixes for that.  Or even write a transient.

bug#56432: run-dig vs dig -- do we need both?

[Lars Ingebrigtsen]

Stefan Kangas <stefan@marxist.se> writes:

> Do we really need both the `dig' and the `run-dig' commands?
>
> Should one of them be obsoleted?  If we want to be able to specify both
> query-type and server, we could consider using double and triple
> prefixes for that.  Or even write a transient.

Looks like `run-dig' should be obsoleted (since it's just a wrapper
around the "dig" program) and has worse output.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#56432: run-dig vs dig -- do we need both?

[Robert Pluim]

>>>>> On Thu, 07 Jul 2022 10:56:55 +0200, Lars Ingebrigtsen <larsi@gnus.org> said:

    Lars> Stefan Kangas <stefan@marxist.se> writes:
    >> Do we really need both the `dig' and the `run-dig' commands?
    >> 
    >> Should one of them be obsoleted?  If we want to be able to specify both
    >> query-type and server, we could consider using double and triple
    >> prefixes for that.  Or even write a transient.

    Lars> Looks like `run-dig' should be obsoleted (since it's just a wrapper
    Lars> around the "dig" program) and has worse output.

For extra credit, add a ʼserverʼ arg to a wrapper around `dns-query',
and obsolete `dig' as well.

Robert
-- 

bug#56432: run-dig vs dig -- do we need both?

[Lars Ingebrigtsen]

Robert Pluim <rpluim@gmail.com> writes:

> For extra credit, add a ʼserverʼ arg to a wrapper around `dns-query',
> and obsolete `dig' as well.

dns.el requires UDP, I think, so it wouldn't work on Windows?  (Nor all
DNS servers support TCP.)

On the other hand, I don't know whether the dig executable is available
on Windows.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#56432: run-dig vs dig -- do we need both?

[Stefan Kangas]

close 56432 29.1
thanks

Robert Pluim <rpluim@gmail.com> writes:

>     Lars> Looks like `run-dig' should be obsoleted (since it's just a wrapper
>     Lars> around the "dig" program) and has worse output.

Thanks, now done (commit 707124d2b9).

> For extra credit, add a ʼserverʼ arg to a wrapper around `dns-query',
> and obsolete `dig' as well.

This part is left as an exercise for the reader.  ;-)

bug#56432: run-dig vs dig -- do we need both?

[Eli Zaretskii]

> Cc: Lars Ingebrigtsen <larsi@gnus.org>, 56432@debbugs.gnu.org
> From: Stefan Kangas <stefan@marxist.se>
> Date: Thu, 7 Jul 2022 11:36:06 +0200
> 
> close 56432 29.1
> thanks
> 
> Robert Pluim <rpluim@gmail.com> writes:
> 
> >     Lars> Looks like `run-dig' should be obsoleted (since it's just a wrapper
> >     Lars> around the "dig" program) and has worse output.
> 
> Thanks, now done (commit 707124d2b9).

Such a hurry...

run-dig seems to be able to guess the host using
net-utils-machine-at-point, something 'dig' doesn't.  So if we want to
tell people top use the latter, IMO we should teach 'dig' that trick.

bug#56432: run-dig vs dig -- do we need both?

[Eli Zaretskii]

> Cc: Stefan Kangas <stefan@marxist.se>, 56432@debbugs.gnu.org
> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Thu, 07 Jul 2022 11:27:52 +0200
> 
> On the other hand, I don't know whether the dig executable is available
> on Windows.

It is, albeit not by default:

  https://phoenixnap.com/kb/dig-windows

bug#56432: run-dig vs dig -- do we need both?

[Stefan Kangas]

Eli Zaretskii <eliz@gnu.org> writes:

> run-dig seems to be able to guess the host using
> net-utils-machine-at-point, something 'dig' doesn't.  So if we want to
> tell people top use the latter, IMO we should teach 'dig' that trick.

net-utils-machine-at-point does not seem very useful to me (it seems
to pick up any word whether or not it looks like a domain name?), but
perhaps we could use thing-at-point instead?

bug#56432: run-dig vs dig -- do we need both?

[Eli Zaretskii]

> From: Stefan Kangas <stefan@marxist.se>
> Date: Thu, 7 Jul 2022 12:47:42 +0200
> Cc: Robert Pluim <rpluim@gmail.com>, Lars Ingebrigtsen <larsi@gnus.org>, 56432@debbugs.gnu.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > run-dig seems to be able to guess the host using
> > net-utils-machine-at-point, something 'dig' doesn't.  So if we want to
> > tell people top use the latter, IMO we should teach 'dig' that trick.
> 
> net-utils-machine-at-point does not seem very useful to me (it seems
> to pick up any word whether or not it looks like a domain name?)

Is it really any worse than not showing any possibly useful default at
all?

> but perhaps we could use thing-at-point instead?

If thing-at-point can guess machines better, sure, why not?  But if it
isn't better, why not use net-utils-machine-at-point?

bug#56432: run-dig vs dig -- do we need both?

[Stefan Kangas]

Eli Zaretskii <eliz@gnu.org> writes:

> > net-utils-machine-at-point does not seem very useful to me (it seems
> > to pick up any word whether or not it looks like a domain name?)
>
> Is it really any worse than not showing any possibly useful default at
> all?

I found it much worse, but this was mostly due to using the obsolete
INITIAL-CONTENTS argument (which meant you had to delete that less
than useful input to type something useful).  So I've now changed
`dig' to use a standard default instead (commit 7b84f9a5d8).

> > but perhaps we could use thing-at-point instead?
>
> If thing-at-point can guess machines better, sure, why not?  But if it
> isn't better, why not use net-utils-machine-at-point?

It seems like the ffap versions are strictly better, so I've replaced
the net-utils-* versions with ffap-* (commit a651c309fd).

bug#56432: run-dig vs dig -- do we need both?

[Robert Pluim]

>>>>> On Thu, 07 Jul 2022 11:27:52 +0200, Lars Ingebrigtsen <larsi@gnus.org> said:

    Lars> Robert Pluim <rpluim@gmail.com> writes:
    >> For extra credit, add a ʼserverʼ arg to a wrapper around `dns-query',
    >> and obsolete `dig' as well.

    Lars> dns.el requires UDP, I think, so it wouldn't work on Windows?  (Nor all
    Lars> DNS servers support TCP.)

It falls back to TCP on windows, so yes, it would not work with the
small minority of servers that donʼt support TCP.

All we need is Someone™ to fix UDP support on Windows.

Robert
--