From: Jason P. <spy...@pe...> - 2007-12-12 19:04:50
|
I guess the core problem is that the python code (eg *execfile*) and spyce code (eg *include.spyce*) seem to use two different working directories. If these could be sync'ed, it would feel smoother. Obviously, this would break too much other code to consider... ah well :) _jason On Dec 11, 2007 3:11 PM, Jonathan Ellis <jon...@ca...> wrote: > For most use cases it's more convenient to treat the path as a "path" > the way the browser would interpret it. That's the only reason it works > the way it does. If I were to revisit this for spyce 2.2 I would > probably do it the way you suggest and just have the user use > util.url2file if they want the docroot-relative path, but in earlier > versions it's enough extra verbosity to import spyceUtil and do that > that it wasn't worth it. > > -Jonathan > > On Tue, 11 Dec 2007 14:44:25 -0500, "Jason Persampieri" > <spy...@pe...> said: > > Howdy folks. Spyce newbie here. > > > > I have noticed an oddity in the *include.spyce* behavior. > > > > If the argument is a relative path - *include.spyce('../include.spy')* - > > everything works fine. > > > > But if I pass an absolute path - > > *include.spyce('/cluster/home/user/public_html/share/include.spy') > > *- it looks like spyce prepends the spyce home directory (which in this > > case > > is the same as the http home) on to the path - * spyceNotFound: > > spyceNotFound exception: could not find > > "/common/www/html/cluster/home/user/public_html/share/include.spy" > > > > *While the absolute path behavior seems consistent with the > > documentation, > > this is incredibly odd. If I specify an absolute path, shouldn't it > > assume > > I'm pointing to exactly what I want? > > > > this whole thing is especially annoying since I need to *execfile *a > > file > > in the same directory (which *only* takes absolute paths seemingly) and > > hence need to keep both a relative and absolute path reference. Piddly > > to > > be sure, but annoying. > > > > That said, Spyce is really rocking my world, and is *exactly* the right > > solution for my project. > > > > Jason Persampieri > > Programmer > > Boston College Bioinformatics Dept. > |