fix issue 427: use v.getConfigType to determine config type in WriteConfig

[mildwonkey]

use the existing getConfigType function instead of figuring it out anew
#427

[CLAassistant]

All committers have signed the CLA.

[mildwonkey]

@spf13 @bep sorry to do this, but - ping! review pretty please

[sgettys]

Is this ever going to make it into master?

[ghost]

I'd like to know when this gets merged as well

[Fuhrmann]

Could this be merged?

[greg-szabo]

I think using getConfigType is not enough. You can build a configuration without defining a filename and then you can try to run SafeWriteConfigAs which again doesn't require to have a filename in your configuration (you are giving the filename as an input parameter to the function).

The problem is that getConfigType will check the filename to get the extension from it. So, in the above-mentioned case, in SafeWriteConfigAs the result of getConfigType will be an empty string: no filename was set in the object, there's no way to figure out it's file extension.

I've tried to figure out why writeConfig is dealing with file extensions and I think the reason is that there's the point where all filenames are in place and the file extension can be figured out. I think the real issue is that the file extension becomes an ultimate priority, even in cases where the developer explicitly set a specific config type using SetConfigType.

In #934 I'm proposing a solution that's making the config type the priority during write, so files with weird extensions (but a properly set ConfigType) can also be saved.

Edit: Also #934 is up-to-date with master. ;) I sure hope that helps with the merge time.