I'm trying to set up my Mercurial repository system to work with multiple subrepos. I've basically followed these instructions to set up the client repo with Mercurial client v1.5 and I'm using HgWebDir to host my multiple projects.

I have an HgWebDir with the following structure:

http://myserver/hg
    fooproj
    mylib

where mylib is some collection of common template library to be consumed by fooproj. The structure of fooproj looks like this:

 fooproj
     doc/
     src/
     .hgignore
     .hgsub
     .hgsubstate

And .hgsub looks like:

src/mylib = http://myserver/hg/mylib

This should work, per my interpretation of the documentation:

The first 'nested' is the path in our working dir, and the second is a URL or path to pull from.

So, let's say I pull down fooproj to my home folder with:

~$ hg clone http://myserver/hg/fooproj foo

Which pulls down the directory structure properly and adds the folder ~/foo/src/mylib which is a local Mercurial repository. This is where the problems begin: the mylib folder is empty aside from the items in .hg. With 2 seconds of investigation, one can see the src/mylib/.hg/hgrc is:

[paths]
default = http://myserver/hg/fooproj/src/mylib

which is completely wrong (attempting a pull of that repo will give a 404 because, well, that URL doesn't make any sense). Logically, the default value should be what I specified in .hgsub or it would get the files from the repository in some way.

None of the Mercurial commands return error codes (aside from a pull from within src/mylib), so it clearly believes that it is behaving properly (and just might be), although this does not seem logical at all.

What am I doing wrong?