Add Spacechem as an humblebundle game

[layus]

No description provided.

[aszlig]

Thank you, after switching to buildGame I'll merge :-)

[layus]

@aszlig I fixed according to your comments.

I had troubles with {setS,s}ourceRoot and fixed some code in buildGame. I am also not too sure about coreutils and uname invocations made by the game.

If this is ok, do not forget to autosquash before merging.

[layus]

@aszlig Wanna have a second look at this one ?

[layus]

Updated, as a new fixed version was released in the meantime.

[layus]

@aszlig Could you have a look at #18, #25 and #21 ?

[layus]

Oh, BTW, the humblebundle downloader does not work anymore. Not sure if this is really important, especially given that the upstream python lib has not been updated since 2015.

[aszlig]

@layus: Hm, one question remains though: You haven't set a sandbox path, so I'd assume the game writes save files in $XDG_DATA_HOME or $XDG_CONFIG_HOME, right? And if yes, it might be a good idea, to set a more specific path.

[aszlig]

@layus: Regarding the HB downloader, I'm already aware of that and will fix it along with the merge of your PRs.

[layus]

Updated the sandbox path. There is a strange access to a pulse "cookie" in CONFIG_HOME/pulse. It gets created by spacechem, but it needs write permission there.

[aszlig]

@layus: While this is just an ugly thing that's not a stopper for merging this PR, can you post a strace -f of the game somewhere so we can figure out why this is is needed?

[layus]

Here it is https://gist.github.com/dab849ee90e7384b45616c07379064e0 :-).

I grep'd 'layus' for brevity.

[layus]

and the same, when "$XDG_CONFIG_HOME/pulse" is removed from the required sandbox paths: https://gist.github.com/3fb6dcb1de51a351d3ee16f49894a5aa

[layus]

https://www.reddit.com/r/spacechem/comments/1qundh/fully_functional_spacechem_on_linux/ has interesting remarks. I have no idea what xclip is for. never encountered the bug.

[aszlig]

@layus: Regarding the pulse cookie, this might be also due to a mixture of 32/64 bit libraries as well, because usually the pulse cookie should be gathered via the X session, but for some reason not included in the strace (as it's only -e trace=%file), it wasn't able to do that.

[aszlig]

https://www.reddit.com/r/spacechem/comments/1qundh/fully_functional_spacechem_on_linux/ has interesting remarks. I have no idea what xclip is for. never encountered the bug.

I guess Spacechem is calling xclip to put data into the user's clipboard. If that's the case, there is monogamePatcher, which can be used to patch the assemblies. It's just a guess and it could also be Mono itself, so I'd double-check with monodis and look for clipboard references.

[layus]

It is a .exe (admitedly, ad 32bit .exe) that runs with mono. It works perfectly well with a 64 bit mono, so I do not get the point of downloading all the 32 bit version of all the dependencies just to autoPatchelfHook patch rgb2theora correctly.

[layus]

It looks like it is working properly with 32bits.

[layus]

I could not find the source that calls xclip, but adding it to path fixes the issue. It's good enough for now.

Also removed the "cookie" sandbox path. Seems not needed for sound to play.