Hi,
i repeated "apt-get update" without getting error messages.
The subsequent "apt-get dist-upgrade" was quite sparse. So i think
the previous one was complete despite the "Err:" messages.
I wrote:
> > In contrast to that update in 2025, some final lines after
> > Reading package lists... Done
> > were missing today:
> > Building dependency tree...
> > Reading state information...
> > 873 packages can be upgraded. Run 'apt list --upgradable' to see them.
David Wright wrote:
> When I update (not upgrade), "Reading package lists... Done"
> is the final line I expect to see.
> [...]
> The lines:
> Reading package lists... Done
> Building dependency tree... Done
> Reading state information... Done
> come from (dist-)upgrade,
Older upgrade logs reveil that in the past, both commands "update" and
"dist-upgrade" reported the three lines which are now missing in
"update". So i think that this double work is avoided now.
> Be aware of the .gpg→.pgp filename change, though this should be
> mitigated by new symlinks in /usr/share/keyrings/.
The system worked after both "dist-upgrade" runs for the narrow use
case of checking whether gcc-15 is still hallucinating about s[n]printf
format overflows like
xorriso/opts_d_h.c:2851:34: warning: ‘%s’ directive writing up to
63519 bytes into a region of size 40960 [-Wformat-overflow=]
2851 | sprintf(xorriso->result_line,"%s\n",tpt);
| ^~
xorriso/opts_d_h.c:2851:4: note: ‘sprintf’ output between 2 and
63521 bytes into a destination of size 40960
2851 | sprintf(xorriso->result_line,"%s\n",tpt);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
where "tpt" is a pointer to a single element of
static char text[][80]={ ... }
(I also tried to change sprintf() to snprintf(..., 1024, ...) with not
more success than the "40960" changing to "1024". It is a misery.)
Eduard Bloch wrote:
> I would report it. This may be not a real bug but the message should
> contain a better explanation of the original cause.
I have now put it on my todo list. (It might last until i get the
problem with firefox-esr and the Debian mailing list archive solved.)
But the real pain is this gcc hallucination for which i'd like to have
a workaround before the next release of xorriso. (It also happens with
libisofs/hfsplus.c with
1411 | sprintf(new_name, "%s_%s", prefix, number);
which is guarded by an if-clause which would prevent the spintf if the
two name components would really cause overflow.)
> That cause may be something simple, like your previous APT cache being
> very old, therefore it has slided out of the patch update time window
> (IIRC the smart patch update are only applicable for a couple of weeks).
> It obviously would have to refetch the complete index files to fix the
> cache. But in any case, the message should be MUCH LESS scary to regular
> users.
... what a drag it is getting old ...
Have a nice day :)
Thomas
