So I feel like I am close to having closed up most of the original issues with the patch. After the conversation with Ted I was feeling good about how to wrap the source indexing so that it’s only applied to a Win32 platform.

The newest patch is here and basically the outstanding issue is just the path to the pdbstr binary. What I’m stuck on is that right now I’m passing a path to pdbstr in my build directory but when this patch is committed it will be somewhere else – which may allow for the setting of the path using global variables in the Makefile that I can’t use on my local debug build.

Some sleuthing should clear that up soon. Perhaps 0.6 will be a commit? I’m setting my sights high for that, and then I could spend some time writing a pdbstr replacement in the last few releases.

Recently I had a conversation with Ted on IRC about how to take steps to make the patch work properly for Win32 platforms.

Here’s the gist of what I’ll be working on for 0.6:

ted: http://lxr.mozilla.org/mozilla/source/Makefile.in#174
ted: that's where symbolstore.py actually gets run
ted: you see how it uses $(MAKE_SYM_STORE_ARGS) ?
ted: up above that line we define those per-platform
ted: like windows gets -c to copy .pdb files
ted: in the main() function in symbolstore.py, we process the cmdline args
ted: so ideally we'd have another flag there
ted: --generate-source-server, or something

That’s my task for now. Off to play with the Makefile.