2008/9/30 frank wang <[EMAIL PROTECTED]>: > I really do not know the difference of debug mode and the pdb debugger. To > me, it seems that pdb is only way to debug the python code. How do the > expert of numpy/python debug their code? Are there any more efficient way to > debug the code in python world? I used to use matlab which comes with a nice > debugger. But now I really want to try the open source software and I found > python/numpy is a nice tool. The only lagging piece of python/numpy > comparing with matlab is the powerful debuggign capability.
I think you will find that the debugger, pdb, is a good analog to matlab's debugging mode. The way I usually debug a piece of code looks something like: [in window A] edit foomodule.py to add a function frobnicate() [in window B, which has a running ipython session] >>> import foomodule >>> reload(foomodule) >>> frobnicate() <some exception happens> >>> %pdb >>> frobnicate() <exception happens again> (pdb) print some_variable 47 [in window A] <wait a minute, some_variable should be 13 at this point...> edit foomodule.py to fix the bug [in window B] >>> reload(foomodule) >>> frobnicate() 21 >>> Alternatively, I write a test_foomodule.py, which contains a suite of unit tests. I run these tests and get dropped into a debugger when one fails, and I try to track it down. The main problem you ran into is that the debugger is not just any old python prompt. It has its own commands. It also, confusingly, will accept python commands, as long as they don't conflict with an internal command. I got into the habit of, when I wanted to run some python thing, writing (pdb) p 17+13 rather than just (pdb) 17+13 If you had simply written (pdb) p where(a>13) everything would have been fine. Anne _______________________________________________ Numpy-discussion mailing list Numpy-discussion@scipy.org http://projects.scipy.org/mailman/listinfo/numpy-discussion
