Francisco Lopes

@mgutz ok thanks, I find the idea of logxi great, the only sad part is about trying some self-contained sample. I'm trying to run tests to check this, but some...

@mgutz ah, I'm sorry, as I got a callstack like the following: ``` 16:15:29.096609 WRN models Could not connect, retrying ... dsn: dbname=testdb in: main(main.go:32) 16:15:29.097692 ERR server Could not...

@mgutz another down point I have about it is that it requires Go 1.3, which isn't available in Ubuntu 14.10 standard repositories. Because of this, at my job, I would...

@mgutz just updated, is it expected to get this callstack? ``` 16:22:56.274622 WRN models Could not connect, retrying ... dsn: dbname=testdb in: main(main.go:32) 16:22:56.275839 ERR server Could not read config...

@mgutz Nice. About closing, this issue in truth is solely to warn about the empty `src` while printing, which seems non-intentional.

@mgutz ok, just one more question. You're using gopkg.in for godo, but for logxi you're using a different kind of internal directory based versioning. And the package name not matching...

@mgutz Interesting! Just when I looked at this I thought about whether two possible improvements would work: - use `logxi.v1` directories, so that `logxi` named package and `import` line matched....

By the way, this pattern is rather different from what I see elsewhere, but if it works, ok, may be a better way out for versioning.

If adopting `.vN` dirs as submodules worked, you would be back to DVCS.

Also, more in line with what you're already doing regarding subdirectory names. Having a packaged named `logxi` instead of `log` for a `logxi/v1` import is much less surprising.