[Baypiggies] Web Crawler/Backend Engineer - San Francisco, CA

[K. Richard Pixley]

Keith Dart wrote:
> But in a language/runtime that has exceptions, like Python, you don't
> really need them. Just raise an exception where you would have the
> "goto", and catch it where you would have the label. I think that is
> even more readable than the goto.
Precisely.

I've also found that in any complex logic where a goto might be called 
for, the complexity itself was effectively a bug both in code 
transparency as well as in performance.  Rewriting such code to avoid 
the use of goto has, in every single case for me, yielded not only more 
transparent code, but also a smaller code foot print and more 
efficiently running code.

Flags are a similar issue.  If you need to create, set, and test flags 
solely in order to maintain your control flow, then that section of code 
almost certainly needs to be redesigned from a higher level, imo.

I'll admit there are functional exceptions.  In assembler.  And 
occasionally those same exceptions can be coded in C.  Things like 
processor bootup code, initial stack construction, the framing code 
around unix processes, process load code, unexec, certain trampolines 
(including those dealing with calls between processor architectures as 
with arm/thumb, implementations of shared libraries, kernel/user space 
transitions, some kinds of coprocessor support, varying or changing call 
semantics, etc), implementations of things like setjmp/longjmp, 
exceptions, continuations, jit emulations like qemu, etc.  But aside 
from academic curiosity or thrill seeking, I don't see much point or 
need for doing so in python.

I once had a friend who thought it would be cool to write a C compiler 
in shell.  To me that's sort of like tearing a Ferrari apart solely with 
a pair of pliers.  While it might be possible, and a challenge, it seems 
to me that one could find similar challenging possibilities which /also/ 
result in useful end products.

I suppose one could always just declare it to be "art".

--rich
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/baypiggies/attachments/20100204/8aa67370/attachment-0001.htm>