On 05/06/12 22:28, Jakub Wilk wrote:
> Is this the desired outcome?
No. AFAICT this is because the new entries are pointing to
/usr/bin/nyancat-server rather than /usr/bin/nyancat which makes
reconf-inetd believe that they are not related.
[background]
/usr/bin/nyancat-server is a symlink to /usr
I installed old nyancat-server, enabled the service, appended "-n" to
its command line, and then upgraded the package. Afterwards
/etc/inetd.conf looked like this:
telnet stream tcp nowait nobody /usr/bin/nyancatnyancat
-t -n
telnet stream tcp6 nowait nobody /usr/bin/n
* Jonathan McCrohan , 2012-06-05, 02:53:
Another version uploaded with a revised postinst which is based on the
updated DEP-9 example.
You don't need to check for existence of /usr/sbin/reconf-inetd. In
postinst configure, it's guaranteed that packages you depend on are
unpacked and configure
On 03/06/12 20:56, Jakub Wilk wrote:
> You wrote:
>
> if [ "fgrep -q -x ..." -o "fgrep -q -x ..." ]; then
>
> This condition is always true (also: not very portable). You want this
> instead:
>
> if fgrep -q -x ... || fgrep -q -x ...; then
Thanks for catching this. Another version uploa
* Jonathan McCrohan , 2012-06-03, 03:12:
I've uploaded a new version to mentors.d.n with changes to the postinst
based on Serafeim's suggestions. I think solves all of the remaining
transition issues.
You wrote:
if [ "fgrep -q -x ..." -o "fgrep -q -x ..." ]; then
This condition is always
On 02/06/12 19:59, Jakub Wilk wrote:
> I think we do need -x (or an equivalent of it). If the user appended
> some options to nyancat-server command-line, we must not remove such entry.
>
> BTW, shouldn't you use --pattern (instead of, or maybe in addition to,
> --multi)?
I've uploaded a new vers
On Sat, Jun 02, 2012 at 07:33:31PM +, Serafeim Zanikolas wrote:
> On Sat, Jun 02, 2012 at 07:03:33PM +, Jakub Wilk wrote:
> > * Jonathan McCrohan , 2012-06-02, 19:27:
> > BTW, shouldn't you use --pattern (instead of, or maybe in addition to,
> > --multi)?
>
> --patern is not implemented f
On Sat, Jun 02, 2012 at 07:03:33PM +, Jakub Wilk wrote:
> * Jonathan McCrohan , 2012-06-02, 19:27:
> >>I would add -x/--line-regexp to the fgrep call.
> >>
> >>Wouldn't it be nice if the postinst also take care of the case when
> >>the user enabled the service?
> >
> >Couldn't both of these is
* Jonathan McCrohan , 2012-06-02, 19:27:
I would add -x/--line-regexp to the fgrep call.
Wouldn't it be nice if the postinst also take care of the case when
the user enabled the service?
Couldn't both of these issues be addressed by removing '## ' from
the grep?
I think we do need -x (or a
On 2 June 2012 18:23, Jakub Wilk wrote:
> But using lt has benefits too: if the user removed old nyancat-server, then
> installing the new one would clean after the buggy postrm.
Thanks for the explanation, seems reasonable.
> I would add -x/--line-regexp to the fgrep call.
>
> Wouldn't it be ni
On Sat, Jun 02, 2012 at 05:27:30PM +, Jakub Wilk wrote [edited]:
[..]
> I would add -x/--line-regexp to the fgrep call.
>
> Wouldn't it be nice if the postinst also take care of the case when the
> user enabled the service?
the line I'm suggesting in dep9 is:
if fgrep -q '^exact entry p
* Jonathan McCrohan , 2012-06-02, 17:42:
Shouldn't it be s/lt/lt-nl/?
What difference does it make? Surely if the package was not previously
installed, the grep match would fail, making the rest of the postinst
no-op.
Well, the line could have been added by user, in which case we shouldn't
t
On 02/06/12 17:19, Jakub Wilk wrote:
> Shouldn't it be s/lt/lt-nl/?
What difference does it make? Surely if the package was not previously
installed, the grep match would fail, making the rest of the postinst no-op.
Jon
--
To UNSUBSCRIBE, email to debian-mentors-requ...@lists.debian.org
with
* Jonathan McCrohan , 2012-06-02, 15:19:
if dpkg --compare-versions "$2" lt "1.0+git20120523.99dc310-1"; then
Shouldn't it be s/lt/lt-nl/?
--
Jakub Wilk
--
To UNSUBSCRIBE, email to debian-mentors-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.deb
On 02/06/12 15:19, Jonathan McCrohan wrote:
> Thanks for the comments; I'll upload a new version which forces a
> reconf-inetd update in a few minutes.
Uploaded.
Jon
--
To UNSUBSCRIBE, email to debian-mentors-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas..
On 02/06/12 12:29, Serafeim Zanikolas wrote:
> In any case, the solution is simple: invoke reconf-inetd right after
> update-inetd in postinst.
That is a simple solution, thanks.
> On a different topic, nyancat/1.0+git20120519.5fe3de9-1 checks for only a
> specific release:
>
> if [ "$2" = "
quoting Jakub Wilk:
> But next reconf-inetd run can happen a month later. (Or never.) In the mean
> time, nyancat server won't be running, will it? Or did I misunderstand
> something?
That's right. I take responsibility for that; I didn't realise that the
postinst script would run after the reconf
* Jonathan McCrohan , 2012-05-31, 23:37:
Why --multi in the update-inetd call?
Because the new reconf-inetd entries are added before the postinst runs.
update-inetd will see multiple telnet entries and alert the user that
multiple servers exist for that service. --multi allows update-inetd to
On 31/05/12 19:04, Jakub Wilk wrote:
> * Jonathan McCrohan , 2012-05-25, 23:57:
>> * Use reconf-inetd to provide nyancat-server configs
>
> Now I realized the old postrm was broken... Ooops.
Yes, I realised that too while I was creating this package, but it will
be fixed by this postinst.
> Why
* Jonathan McCrohan , 2012-05-25, 23:57:
* Use reconf-inetd to provide nyancat-server configs
Now I realized the old postrm was broken... Ooops.
Why --multi in the update-inetd call?
--
Jakub Wilk
--
To UNSUBSCRIBE, email to debian-mentors-requ...@lists.debian.org
with a subject of "unsub
[I've uploaded a new version 1.0+git20120523.99dc310-1 [1] to
mentors.d.o]
On 25/05/12 11:34, Jakub Wilk wrote:
> I'd rather not add the override, but ignore the tag for the time being.
> hardening-no-stackprotector will be disabled in the next lintian release.
I've removed this override.
> You
* Jonathan McCrohan , 2012-05-25, 01:24:
- Add hardening-no-stackprotector override; false detection
I'd rather not add the override, but ignore the tag for the time being.
hardening-no-stackprotector will be disabled in the next lintian
release.
(Also: s/override/lintian override/. Othe
[I've uploaded version 1.0+git20120523.99dc310-1 [1] to mentors.d.o]
On 24/05/12 16:09, Jakub Wilk wrote:
> Well, I have it installed. Is that enough? :)
Fair enough. This is fixed in 1.0+git20120523.99dc310-1.
> I'm confused. What "NEWS page"?
A NEWS.Debian file. Section 6.3.4 of the Debian De
* Jonathan McCrohan , 2012-05-23, 11:48:
I think you need some extra code in postinst to deal with this
transition. See 'Transition of "Depends: update-inetd" packages'
section of DEP-9.
I purposely haven't added postinst code because according to popcon
[1], nobody has installed the package.
On 23 May 2012 11:27, Jakub Wilk wrote:
> I think you need some extra code in postinst to deal with this transition.
> See 'Transition of "Depends: update-inetd" packages' section of DEP-9.
I purposely haven't added postinst code because according to popcon
[1], nobody has installed the package.
* Jonathan McCrohan , 2012-05-20, 03:20:
nyancat (1.0+git20120519.5fe3de9-1) unstable; urgency=low
* New upstream release
- Fixes buildflags being incorrectly passed. Allows build hardening flags.
* Switch to debhelper v9
OK.
* Use reconf-inetd to provide nyancat-server configs
I thi
Package: sponsorship-requests
Severity: normal
Dear mentors,
I am looking for a sponsor for my package "nyancat"
It builds those binary packages:
nyancat - Animated terminal Nyancat
nyancat-server - Animated terminal Nyancat server configurations
To access further information about this packag
27 matches
Mail list logo
