Ian Holmes

There are a few situations I can imagine where having the target metadata associated with the Makefile location (or the current working directory), rather than the target location, would be...

I'm still kind of hoping that an elegant solution will present itself... e.g. if @cmungall jumps in... I can certainly mitigate it by - improving the error message - adding...

Also, if we allow for switchable behavior, there are other options... for example the location of metadata directories (in CWD vs in target dir) could be a user-changeable option... seems...

I guess there are two possibilities: the test is now broken, or it was broken before and you just fixed it, but the "what it's meant to look like" file...

@cmungall The purpose of this test (test-159) was to check that biomake would recompute the MD5 hash if the file in which the hash was cached had an older timestamp...

The more I think about it, the more I think that rebuilding the target file actually *is* the correct behavior if its MD5 hash file turns out to be stale....

Only admin users should be able to lock symbols in the public namespace Private namespace available at `/~username/symbol`

Just needs save

pretty much there as of 440ce72 but not tested or hooked up

If we force everything to be a Markov chain, then it's memoryless and you don't need time on last page...