Bug#1086770: vcmi: Very unstable package

[Alex Henry]

Hi Johannes! Thank you for your work on Debian and VCMI!

I was not playing with anything other than the internal VCMI packages that
come installed by default. I tried to play one or two random-scenarios but
had to give up mid-way through.

I do believe the instability of the package to be a series of different
bugs, not a single underlying issue as the game would immediately crash in
response to specific actions (which I've failed to document, both in-side
and out-side of combat).

I cannot promise that I'll have the time to test any further but if you
remind me when you upload 1.5.7, I'll try to find time if I'm able.

Here's a couple wild-guesses:

#1: Could be a problem with a specific version of a library. You could try
ensuring the package depends on the latest versions of all packages. I did
notice I had to update one or two packages to install 1.5.7 from the PPA.
As I said earlier, my system is otherwise fairly up-to-date, I believe.
#2: Could have been some sort of compatibility issue with my previous
configuration files? If I'm able to test the next version, I'll have to
remember to do a clean install. I'm not even sure if I had any previous
files but wouldn't hurt to make sure, next time around.

I'll suggest again you consider bumping this bug-report severity so the
package won't migrate to stable. I'm sure users would prefer an out-dated
but playable version of the engine than be stuck with a version that
crashes every 10-20 minutes and loses progress (at least for some users)
for an entire stable cycle.

On Wed, 6 Nov 2024 at 07:10, Johannes Schauer Marin Rodrigues <
josch at debian.org> wrote:

> Hi Alex,
>
> thank you for your elaborate bug report.
>
> Quoting Alex Henry (2024-11-05 18:53:13)
> > I have played VCMI for about an hour yesterday and have run into numerous
> > crashes-to-desktop during basic, normal game-play. These include a
> reproducible
> > crash where pressing the space-bar during combat will crash-to-desktop
> with a
> > 100% chance, every time.
>
> that's this bug: https://github.com/vcmi/vcmi/issues/4067
>
> This should be fixed in vcmi 1.5.3.
>
> Indeed I must admit that I wasn't aware of this hotkey and always used my
> mouse
> to let units wait.
>
> > The other crashes happened frequently too, every 10 or 20 minutes or so,
> also
> > usually after some input-action, suggesting a thorough lack of testing
> and
> > polish.
>
> Are you playing with any mods enabled? I have been playing the campaign
> yesterday for over an hour without any crashes.
>
> > Considering every crash causes anywhere from 5 to 10 or more minutes of
> data
> > loss, I'll argue that this package is unusable for anyone but the most
> > patient of users who don't mind having 50% of their progress lost
> frequently
> > while playing a video-game past the 1.0 mile-stone (which won't be many).
> >
> > I have played earlier versions of VCMI and it was not anywhere as
> unstable as
> > this and as such I'd recommend the maintainers block this package from
> > reaching stable. I have tried version 1.5.7 too from the project's
> official
> > "stable" PPA and also ran into a crash within minutes of launching the
> game.
>
> I'm currently preparing an upload of version 1.5.7 to unstable. I'll test
> that
> as well and see if I can reproduce your observations.
>
> > I do apologize to the maintainers for not providing specific logs and
> > detailed reports of each individual bug but:
> >
> > 1. I have not collected every crash-log and it would be more
> time-consuming to
> > do so than I could do right now, considering the sheer number of crashes.
> > 2. The individual issues are not the problem: I could continue playing,
> just by
> > avoiding pressing space during combat, for example. The problem is the
> overall
> > lack of stability in the entire package that renders this pretty much
> > unplayable, in my opinion, to the extent I feel the need to write this.
> >
> > While I have not tested it recently, I do believe the 1.1 version in
> stable
> > does not suffer from this trouble-some lack of polish and I would
> suggest that
> > migration be blocked until at least 1.5.8 is released in the future and
> > properly tested. Consider making this a "serious" bug, if applicable.
>
> I think it would be useful to look at some of your crash logs, even better
> with
> the dbg package installed and then with gdb. As I didn't run into any crash
> (other than the spacebar issue) I'm puzzled what the underlying issue of
> our
> differing experiences could be.
>
> After I uploaded 1.5.7 to unstable, would you have the time to upload some
> crash logs with backtraces as issues to upstream? Feel free to @-mention
> me.
> I'm "josch" on github.
>
> Thanks!
>
> cheers, josch