The program can not run

[summer]

After the 1.31 version of the installer can not run.

[Iris]

We don’t have a version 1.31. Also, perhaps you could be more specific? I’m assuming this is on Windows since you said ‘installer’. Do you get any error messages? Could you take a screenshot if that’s the case?

(Moving this to Technical Support.)

[summer]

We don’t have a version 1.31. Also, perhaps you could be more specific? I’m assuming this is on Windows since you said ‘installer’. Do you get any error messages? Could you take a screenshot if that’s the case?

(Moving this to Technical Support.)

Sorry, 1.13.1 version, the installation is complete, click wesnoth.exe invalid.

[Anonymissimus]

Please post stderr.txt:

On Windows, locate the file stderr.txt. If you have installed Wesnoth in the default location, that would be C:\Program Files\Wesnoth\stderr.txt — “Program Files” might be translated to your system’s native language. Search for the file if you can’t find it. Windows Vista and Windows 7 users should read this thread for further information.

[Iris]

I’m more interested in a screenshot of the actual error message, because if it’s what I think it is there’ll be no stderr.txt.

[summer]

Please post stderr.txt:

On Windows, locate the file stderr.txt. If you have installed Wesnoth in the default location, that would be C:\Program Files\Wesnoth\stderr.txt — “Program Files” might be translated to your system’s native language. Search for the file if you can’t find it. Windows Vista and Windows 7 users should read this thread for further information.

stderr.txt:
Battle for Wesnoth v1.13.1
Started on Fri Jul 02 09:01:54 2015

Automatically found a possible data directory at D:/Battle for Wesnoth
Caught general exception:
boost::random_device: Could not find provider name Cryptopraphic Service Provider Microsoft Base Cryptographic Provider v1.0:

[Iris]

Yeah, that’s definitely not what I was expecting when you said “wesnoth.exe invalid” in response to my question about error messages.

And what version of Windows is this again?

[iceiceice]

Hmm... I think this could be fixed by putting all the random device stuff in a try / catch block in here:

https://github.com/wesnoth/wesnoth/blob ... ng.cpp#L54

Didn't know that thing could throw...

[Iris]

Hmm... I think this could be fixed by putting all the random device stuff in a try / catch block in here:

https://github.com/wesnoth/wesnoth/blob ... ng.cpp#L54

Didn't know that thing could throw...

Isn’t it more important to make sure the runtime environment can supply this service, otherwise it defeats the point of adding this in 1.13.0? Also note that it works for me on both Windows 7 and 8.1.

[iceiceice]

If it's throwing the exception, it means that it can't provide the service. Setting up time(NULL) as a fallback is perfectly reasonable in that case I think.

I don't think there's going to be an alternative like "try to build boost random device with different options to recover from the failure", the point is that they want to provide a simple cross platform interface, so if there was some possibility like that i think they would have built it in. It's possible that there's something wrong in the boost code for whatever version of windows he's on.

Edit:
You could signal a warning when the fallback generator is used?

Note that what is being selected here is *not* the rng algorithm, which will always be the mt19937 thing. This is just the part that is generating *seeds* for that generator. Random device is the preferred way to generate these seeds in boost docs, it has some advantages over using time(NULL). But I mean unless you are running the mp server these advantages are pretty unimportant.

[summer]

Yeah, that’s definitely not what I was expecting when you said “wesnoth.exe invalid” in response to my question about error messages.

And what version of Windows is this again?

win7 X64