version checking uses wrong 'version' command

[nachog00]

The current version checking implementation for binaries assumes all binaries support the --version api.
Some of them actually work with a version command/first arg.

[ala-mode]

What's in place now is a workaround for this issue (looking at stderr as well as stdout), which is with two of the binaries we're featuring.

lightwalletd will likely not change to support --version.
zaino already does in its newest dev version after a follow up to this issue. However, the binaries we are currently dealing are not up to date with this.

I would assign this low priority, though when we update the binaries to new versions, we should then use zaino's --version output.

[nachog00]

What's in place now is a workaround for this issue (looking at stderr as well as stdout), which is with two of the binaries we're featuring.

lightwalletd will likely not change to support --version. zaino already does in its newest dev version after a follow up to this issue. However, the binaries we are currently dealing are not up to date with this.

I would assign this low priority, though when we update the binaries to new versions, we should then use zaino's --version output.

If i understand correctly, this counts on the actual version string to come up on either stdout OR stderr, right?

I don't know how strong that assumption is...

An alternative would be to explicitly define each binary's version command then use that. If it changes for any of them in future versions, we can update it to. It could be part of the refactor #27

[ala-mode]

If i understand correctly, this counts on the actual version string to come up on either stdout OR stderr, right?

I don't know how strong that assumption is...

I think that assumption is pretty good! You can see it work. I agree it's not elegant though.

An alternative would be to explicitly define each binary's version command then use that. If it changes for any of them in future versions, we can update it to. It could be part of the refactor #27

Yes, then instead of handling two output types (stdout and stderr, for which there is a working system in place) we're handling two input flags (--version and version, for which there is no working system in place).

Improvement?

I think this is low priority.