Bug#991778: dlint: Dlint fails to find version number of dig

[Paul Gevers]

Control: tags -1 bullseye-ignore

Hi Patrik, Florian,

On Sun, 01 Aug 2021 19:00:34 +0200 Patrik Schindler <p...@pocnet.net> wrote:
> On my Debian Buster system, dig fails to work with

On my Debian bullseye system, this works as expected, see below. So
apparently it's either only broken in buster, or it's not always broken.
As it's too late to fix this for the upcoming release of bullseye in two
weeks, I mark this bug as bullseye-ignore (Release Team member hat on).
If the package is broken in buster only, that may be fixed via a point
release update.

Paul

paul@mulciber ~ $ dlint debian.org
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
WARNING: tempfile is deprecated; consider using mktemp instead.
;; dlint version 1.4.0, Copyright (C) 1998 Paul A. Balyoz <p...@domtools.com>
;; Dlint comes with ABSOLUTELY NO WARRANTY.
;; This is free software, and you are welcome to redistribute it
;; under certain conditions. Type 'man dlint' for details.
;; command line: /usr/bin/dlint debian.org
;; flags: normal-domain recursive.
;; using dig version 9.16.15
;; run starting: di 3 aug 2021 19:58:00 CEST
;; ============================================================
;; Now linting debian.org
;; Checking serial numbers per nameserver
;; 2021080326 sec1.rcode0.net.
;; 2021080326 nsp.dnsnode.net.
;; 2021080326 sec2.rcode0.net.
;; 2021080326 dns4.easydns.info.
;; All nameservers agree on the serial number.
;; Now caching whole zone (this could take a minute)
;; trying nameserver dns4.easydns.info.
ERROR: no A records found.
;; no subzones found below debian.org, so no recursion will take place.
;; ============================================================
;; dlint of debian.org run ending with errors.
;; run ending: di 3 aug 2021 19:58:01 CEST

[Patrik Schindler]

Hello Paul,

sorry for letting you wait for so long. Had a lot of other, more important duties at hand.

Meanwhile I'm at Bullseye, and the issue persists.


> Am 04.09.2021 um 06:58 schrieb Paul Wise <pa...@debian.org>:
>
> Control: severity -1 important
> Control: tags -1 + unreproducible

>
> On Sun, 01 Aug 2021 19:00:34 +0200 Patrik Schindler wrote:
>
>> On my Debian Buster system, dig fails to work with
>>

>> ;; This program requires DiG version 2.1 or newer, which I cannot find.
>
> I can't reproduce this in a Debian buster chroot.
>
> Could you investigate more closely why it fails on your system?

I'm using a .digrc to minimize output for my "ordinary" user:

+ret=2 +noauthority +noadditional +noquestion +nostats +nocmd

For root (with no .digrc), the stock command still works.

>> Checking on this, I saw that the "ver" call in line 109 doesn't output a
>> version number anymore. So I changed that to dig -v.
>
> The command there still works for me in buster:
>
> $ dig localhost any | grep DiG | head -1 | sed -e 's/.*DiG \([0-9.]*\).*/\1/'
> 9.11.5

Yes, unless you make dig less talkative with +nocmd.

>> The following replacement for line 109 works for me.
>>
>> ver=`dig -v 2>&1 | grep DiG | head -1 | sed -e 's/^DiG \([0-9.]\+\).*$/\1/'`
>
> This seems like a much more correct way to check the version, so it should be used instead anyway.

Thanks for confirming my opinion.

:wq! PoC

[Paul Wise]

Control: tags -1 - unreproducible + confirmed patch

On Fri, 2022-07-22 at 14:00 +0200, Patrik Schindler wrote:

> I'm using a .digrc to minimize output for my "ordinary" user:
>
> +ret=2 +noauthority +noadditional +noquestion +nostats +nocmd
>
> For root (with no .digrc), the stock command still works.

After copying your dig config I can now reproduce the error.

> Yes, unless you make dig less talkative with +nocmd.

That looks like the cause of this problem indeed.

This only shows up in the non-default configuration, removing the
non-default option works around the issue and users have an easy change
they can apply to fix the issue while keeping the non-default option.

Consequently I think that important is the right severity and that the
bug should not cause removal of dlint from Debian testing (bookworm).

I hope the dlint maintainer (CCed) can fix this, but they don't seem to
be very active in Debian. There is also no upstream for the package.
So the bug isn't likely to get fixed any time soon. If you have time to
maintain the package, salvaging it might be the best way forward.
Otherwise please apply the fix locally for now.

--
bye,
pabs

https://wiki.debian.org/PaulWise

[Patrik Schindler]

Hello Paul,

Am 23.07.2022 um 02:06 schrieb Paul Wise <pa...@debian.org>:

>> Yes, unless you make dig less talkative with +nocmd.
> That looks like the cause of this problem indeed.

Thanks for confirmation!

> So the bug isn't likely to get fixed any time soon. If you have time to maintain the package, salvaging it might be the best way forward.

I'll think about it. Seems I won't inherit a huge list of unresolvable bugs. My decision might be based upon the availability of a concise "what it takes to be a package maintainer" documentation.

> Otherwise please apply the fix locally for now.

I already did. :-)

Btw., I apparently created a second bug report, this one (991778) for the same issue. Browsing the still open bugs list it appears, I've created a bug report for the same issue a decade before, 619936. And already provided a similar solution. Which didn't make it into the releases in between.

Maybe close one of them, marked as dupe?

:wq! PoC

[Paul Wise]

Control: forcemerge 991778 619936

On Wed, 2022-07-27 at 00:20 +0200, Patrik Schindler wrote:

> I'll think about it. Seems I won't inherit a huge list of
> unresolvable bugs. My decision might be based upon the availability
> of a concise "what it takes to be a package maintainer"
> documentation.

The basic process is that you upload package changes to mentors.d.n and
then search for a sponsor, who inspects the package and uploads it to
Debian. The maintainer should respond to bug reports and other queries.
Since this package also has no upstream project, the maintainer should
also try to fix bugs that are reported rather than forwarding them
upstream. Since dlint is also in several different distros, it might be
worth creating a new upstream project and using that as the basis of
the Debian package, then getting other distros to use it too, this
would be optional of course though. You'll need a mentors.d.n account
and since the dlint package is stored in git, you will also need an
account on the Debian Salsa service. Since the git repo is in Florian's
namespace, that would also need to be moved to the debian namespace.
Perhaps if Florian returns he could become your sponsor for this :)

https://mentors.debian.net/intro-maintainers/
https://tracker.debian.org/pkg/dlint
https://repology.org/project/dlint/packages
https://wiki.debian.org/Salsa

> Btw., I apparently created a second bug report, this one (991778) for
> the same issue. Browsing the still open bugs list  it appears, I've
> created a bug report for the same issue a decade before, 619936.

In future, please do check for existing bugs before filing new ones,
if you use reportbug then it will show you the list of open bugs.

> And already provided a similar solution.
> Which didn't make it into the releases in between.

I guess Florian didn't have time to look at all the bugs when he made
the last dlint update in 2019.

> Maybe close one of them, marked as dupe?

We can simply merge the bugs, which keeps it open but hides it when the
bug list is set to hide merged bugs. Done using the command above.

https://www.debian.org/Bugs/Reporting#control
https://www.debian.org/Bugs/server-control#forcemerge