Hi Martin,
your package buffer came up as a candidate for the Bug of the Day[1]
today. Thus I had a *short* look on it. I've (quickly!) created a repository for the package[2] for your comfort (its far from finished).
If you agree the Salvage team might update the packaging in the next
couple of days and fix the bugs in CC.
We realise that this software is orphaned long ago and somehow in maintenance mode. Do you think it is OK to maintain the package in Debian team so we can just upload or do you want to do the final upload yourself? It would be great
if you could raise your opinion in the next couple of days. We could also do some NMU to delayed=15 which was done in similar cases over last time.
Kind regards and thank you for maintaining the package over the years
Andreas.
[1] https://salsa.debian.org/tille/tiny_qa_tools/-/wikis/Tiny-QA-tasks#bug-of-the-day
[2] https://salsa.debian.org/debian/buffer
--
https://fam-tille.de
back from skiing, sorry for the late reply ;-)
I've not had a close look yet at the salsa repo, but in general this all sounds reasonable. I'm fine with team maintenance in the Debian group, feel free to go ahead.
Regarding buffer in general: IMO, it's really time to replace it with something else. It relies on SysV shared memory which no longer is state of the art and thus is quite limited these days. But I no longer use buffer myself, so I've not actively searched for better/more modern alternatives.
So until we can recommend one, it probably make sense to keep buffer alive (or rather un-dead ;-)
Inside the Tiny Tasks Matrix channel наб asked:
"isn't mbuffer the same as buffer but better?"
According to your insight into buffer would you consider this some
sensible equivalent? If so I'd volunteer to file some RoQA bug to ask ftpmaster for removal of buffer.
On Wed, Mar 05, 2025 at 10:44:45AM +0100, Andreas Tille wrote:Ideally we could provide buffer with mbuffer, but this may require a wrapper; just glancing at the manuals I see a different default size and no -S;
Inside the Tiny Tasks Matrix channel наб asked:
"isn't mbuffer the same as buffer but better?"
So I'm all for simply dropping buffer and recommending mbuffer instead (is that important enough for the "Noteworthy obsolete packages" section in the release notes?).
Ideally we could provide buffer with mbuffer, but this may require a wrapper; just glancing at the manuals I see a different default size and no -S;
this could just be as simple as a getopt(1) loop however.
...actually mbuffer's default seems to be -S $blocksize
(with a different output format, which is very busy);
there's -q to disable it, but no option to change the logged increments.
And there's no way AFAICT to disable config file loading.
So no dice without significant mbuffer patching.
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 481 |
Nodes: | 16 (2 / 14) |
Uptime: | 11:53:48 |
Calls: | 9,540 |
Calls today: | 8 |
Files: | 13,653 |
Messages: | 6,139,333 |
Posted today: | 1 |