Re: [PATCH proxmox-mini-journalreader 1/2] add -B parameter to lists boots

[Maximiliano Sandoval <m.sandoval@proxmox.com>]

Maximiliano Sandoval <m.sandoval@proxmox.com> writes:

> The parameter will print all boots, in practice it should be piped into tail.
>
> Sample output:
>
> 2026-07-27 08:28	2026-07-27 15:56	3e3813556d53477f9a139a7cb2f3e8f2	7.0.14-6-pve
> 2026-07-27 15:57	2026-07-27 17:03	5a0324e1ac3f4f24907a21ab81b67384	7.0.14-7-pve
> 2026-07-28 08:26	2026-07-28 17:07	e62bc71e85274f21a1bfb954be38c071	7.0.14-7-pve
> 2026-07-29 08:43	2026-07-29 08:48	7f668e4499644ea1b8339d6b8c29d53d	7.0.14-7-pve
> 2026-07-29 08:49	2026-07-29 17:09	9ff5eb6f0485466383e0b89512f937c9	7.0.14-8-pve
>
> glib is added as a dependency for simplicity of memory management.
>
> Signed-off-by: Maximiliano Sandoval <m.sandoval@proxmox.com>
> ---
>
> Notes:
>     We usually require the output from `last` or similar in support to query the
>     previosly booted kernel versions. We add this helper to avoid adding a
>     dependency on either last or wtmpdb on the ISO images.
>     
>     One awkward thing is that doing sd_journal_seek_tail while inside a
>     sd_journal_query_unique results in a segmentation fault, hence we iterate twice.
>     
>     This seems quick enough for ~400 boots:
>     
>     time sudo ./proxmox-mini-journalreader -B | wc -l
>     417
>     
>     real	0m0,143s
>     user	0m0,004s
>     sys	0m0,004s
>     
>     Open questions:
>     
>     Should it list the latest boots first? It amounts to reversing the sign on
>     boot_info_cmp.
>
> [..]

Found a couple of very minor issues, will send a v2 once there is
feedback. v2 does not need to open a second journal, but still needs the
double iteration.

-- 
Maximiliano