Bug#1030857: transmission: New 4.0 version is available

[Tim Sattarov]

Source: transmission
Version: New 4.0 version is available
Severity: normal
Tags: upstream
X-Debbugs-Cc: sti...@gmail.com

Dear Maintainer,


Transmission team has published new version that has many improvements and
fixes.

https://github.com/transmission/transmission/releases/tag/4.0.0

Please consider packaging it.

Thank you


-- System Information:
Debian Release: bookworm/sid
APT prefers testing
APT policy: (500, 'testing'), (50, 'unstable')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 6.1.0-3-amd64 (SMP w/16 CPU threads; PREEMPT)
Locale: LANG=en_CA.UTF-8, LC_CTYPE=en_CA.UTF-8 (charmap=UTF-8), LANGUAGE=en_CA:en
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled

[Leo Antunes]

Hi,

Just FYI: I have done some work in the salsa repo[0], but there are still a few kinks to iron out before we can ship it. It builds, but my debhelper-foo is a bit rusty :)

If anyone wants to jump in and finish it up, I wouldn't complain!

Cheers
Leo

[0] https://salsa.debian.org/debian/transmission

[Barak A. Pearlmutter]

Okay, well...

> Just FYI: I have done some work in the salsa repo[0], but there are still a few kinks to iron out before we can ship it. It builds, but my debhelper-foo is a bit rusty :)

> If anyone wants to jump in and finish it up, I wouldn't complain!

Don't complain, because I did a bit more work, on a fork of that repo,
salsa.debian.org/bap/transmission

I *think* it's good to go, although someone should give it a bit of a
test first. I only use the daemon, so I'm not in a good position for
it.

[Barak A. Pearlmutter]

Testing the 4.0.1 daemon.

Upgrade seems okay, although there's some straggler ghost file
/etc/init/itransmission-daemon.conf

The upgraded daemon invalidates all downloaded data and wants to
verify them, which is a local operation, but is still taking forever.
I think that's supposed to be a feature not a bug.

If someone were to test the clients that would be great!

[Leo Antunes]

Hi,

On Fri, 24 Feb 2023 21:28:18 +0000 "Barak A. Pearlmutter" wrote:
> The upgraded daemon invalidates all downloaded data and wants to
> verify them, which is a local operation, but is still taking forever.
> I think that's supposed to be a feature not a bug.

Yeah, that makes sense.

> If someone were to test the clients that would be great!

Unless morph vetoes it, I'd merge your 3 salsa MRs [0][1][2] and if no obvious problems pop up during testing I'd just upload it. WDYT?


Cheers
Leo Antunes


[0] https://salsa.debian.org/debian/transmission/-/merge_requests/11
[1] https://salsa.debian.org/debian/transmission/-/merge_requests/12
[2] https://salsa.debian.org/debian/transmission/-/merge_requests/13

[Sandro Tosi]

> Unless morph vetoes it, I'd merge your 3 salsa MRs [0][1][2] and if no obvious problems pop up during testing I'd just upload it. WDYT?

no issues here

--
Sandro "morph" Tosi
My website: http://sandrotosi.me/
Me at Debian: http://wiki.debian.org/SandroTosi
Twitter: https://twitter.com/sandrotosi

[Barak A. Pearlmutter]

Okay, I installed the package generated by 16c2e55a0 in my repo, which
is a 4.0.1-1 candidate. It seems to work okay EXCEPT ...

(a) is kicks out messages like this

Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)
Feb 26 21:49:35 sweat transmission-daemon[1174]: [2023-02-26
21:49:35.396] ERR torrent-metainfo.cc:630 no bencoded data to parse
(84) (./libtransmission/torrent-metainfo.cc:630)

And, it is refusing to verify a whole bunch of stuff that to my
knowledge was already downloaded just fine under 3.x. It just lists
them as 0%. Hitting "verify" on them does nothing, nor does resume or
re-announce. Some of them list as downloading; of those, sometimes one
wakes up and verifies and jumps to seeding.

No idea what's going on.

Not sure this is a sufficient show-stopper to preclude uploading this
version, but thought I'd mention it.

Looking at the post-release upstream commits, I suspect some may
address this. I could try cherry-picking some of those commits, or
maybe just merging the development tip, and seeing if that fixes the
problem. Or, maybe we should wait for 4.0.2 which might address this
properly? Is anyone in touch with upstream? It might help to get their
take, find out if they're going to do a point release fixing this
stuff sometime soon.

[Barak A. Pearlmutter]

Okay, I cherry-picked upstream commits 487cc27e1..d21a3b622, the
endpoint being the current upstream/main, and built, and installed,
and it seems to solve this problem. The "no bencoded data to parse"
messages are gone. And things verify upon request, with most of them
succeeding. A few failed to verify even though they are absolutely
downloaded; these are all single file torrents, instead of a directory
containing files. So that's a clue as to the bug, I suppose.

Anyway, this issue does seem at least mostly fixed upstream post-release.

[Leo Antunes]

Hi,

------- Original Message -------

On Thursday, March 2nd, 2023 at 01:39, Barak A. Pearlmutter <ba...@pearlmutter.net> wrote:
> Okay, I cherry-picked upstream commits 487cc27e1..d21a3b622, the
> endpoint being the current upstream/main, and built, and installed,
> and it seems to solve this problem. The "no bencoded data to parse"
> messages are gone. And things verify upon request, with most of them
> succeeding.

Since we missed the freeze anyway, I'd suggest we give upstream a couple more days for a release including the fixes?

> A few failed to verify even though they are absolutely
> downloaded; these are all single file torrents, instead of a directory
> containing files. So that's a clue as to the bug, I suppose.

What does "fail to verify" mean? Does it invalidate the whole download? Or are they stuck in some state?
My test upgrade triggered verifications for all existing torrents, but once verifications were done, all seemed good.


Cheers
Leo Antunes

[Barak A. Pearlmutter]

For me it triggered verification of all existing torrents, but some of
them which are actually 100% downloaded show as 0% even after
verification. These are (after the cherry picking etc) only ones with
a single file download, rather than a directory. And only some of
them. Some of the ones that got in that state proceeded to download
again successfully. But some could not find any peers and are stuck at
0%, even though I can see that they are actually 100% downloaded. For
ones that proceeded to download again, I *think*, but am not sure,
that after connecting to a peer they verified correctly instead of
actually re-downloading.

Anyway, I think we might be able to squeeze transmission through the
freeze if we upload chop-chop! I thought it was soft now, and didn't
apply to "leaf" packages like this.

[Barak A. Pearlmutter]

Leo,

Thanks for uploading 4.0.1-1. Good idea to disable libtransmission-dev for now.

I did a bit of testing, and it seems to get a bunch of those old "no
bencoded data to parse" errors, and a whole bunch of fully downloaded
torrents showed up as 0%. So I cherry-picked to the tip of
upstream/main (as of yesterday) above your branch, see branch
cherry-pick-devel on https://salsa.debian.org/bap/transmission, and
that fixed *everything*! All fully-downloaded torrents verify
completely and show as 100% downloaded. Yes, even all the single-file
ones. And no more "no bencoded data to parse" errors. If you look at
the commits, there are a bunch of seemingly-relevant bug fixes. And
also some minor stuff.

Anyway, I think it would be a service to our users to not have them
upgrade and suddenly half their complete torrents show up as 0% and
they're getting screenfulls of "bencoded" errors about torrents that
used to work fine. So I think it might be a good idea to upload a
version patched with these upstream fixes, or at least some coherent
subset of them.

Really upstream should release 4.0.2 with these fixes right away,
since 4.0.1 is, technically speaking, what we in the biz call
"broken". But I digress.

Cheers,

--Barak.

[Sandro Tosi]

> Really upstream should release 4.0.2 with these fixes right away,
> since 4.0.1 is, technically speaking, what we in the biz call
> "broken". But I digress.

upstream released 4.0.2:
https://github.com/transmission/transmission/releases/tag/4.0.2

[Leo Antunes]

I'm on it.
But AFAIUI, the only chance of getting it through the freeze would be some autopkgtests[0] :/

Regards,

[0] https://release.debian.org/testing/freeze_policy.html#hard


------- Original Message -------

[Barak A. Pearlmutter]

I would say that marking lots of 100% downloaded torrent as 0%, and
refusing to re-verify them, would count as a severity: important bug,
and hence allow 4.0.2 to get into the release. The release team wants
a high-quality release just as much as we do, and transmission is a
leaf package and therefore they might be a bit more lenient.

Anyway, I've just prepared a 4.0.2-1 release candidate,
https://salsa.debian.org/bap/transmission branch "master" with
"upstream" and "pristine-tar" branches updated as well, and new tag
"upstream/4.0.2". I've tested it, and it solves the "fully downloaded
torrent sticks as 0% and refuses to verify" problem, and seems
otherwise functional.

[Leo Antunes]

hi,

------- Original Message -------

On Friday, March 17th, 2023 at 21:51, Barak A. Pearlmutter <ba...@pearlmutter.net> wrote:

> I would say that marking lots of 100% downloaded torrent as 0%, and
> refusing to re-verify them, would count as a severity: important bug,
> and hence allow 4.0.2 to get into the release. The release team wants
> a high-quality release just as much as we do, and transmission is a
> leaf package and therefore they might be a bit more lenient.

Testing still has 3.00, so I'm afraid bugs in 4.0.1 are probably not an argument for migration.
But who knows, maybe 🤷

> Anyway, I've just prepared a 4.0.2-1 release candidate,
> https://salsa.debian.org/bap/transmission branch "master" with
> "upstream" and "pristine-tar" branches updated as well, and new tag
> "upstream/4.0.2". I've tested it, and it solves the "fully downloaded
> torrent sticks as 0% and refuses to verify" problem, and seems
> otherwise functional.

4.0.2 is already uploaded ;)
https://tracker.debian.org/news/1427244/accepted-transmission-402-1-source-into-unstable/


Regards,
Leo

[Barak A. Pearlmutter]

cool.
force pushed yours to my repo, and rebased some yak shaving onto it