Re: [fetchmail-devel] Some PATRICIA numbers

[Matthias A. <mat...@gm...>]

Am 18.05.2010, 19:48 Uhr, schrieb Rainer Weikusat:

> "Matthias Andree" <mat...@gm...> writes:
>> Am 16.05.2010, 23:33 Uhr, schrieb Rainer Weikusat:
>>> The trie implementation is meanwhile complete
>
> [...]
>
>> I'm really looking forward to your code :-)
>
> I hope that I've now also found the 'obvious' optimizations (even with
> 'full' tail pointers, save_str is the third most expensive operation
> accomplished by fetchmail, so increasing insert times potentially
> matters a lot). I am going to start with integrating the uidl db
> implementation I have with the POP3 code now. If this was ok, I would
> like to send an 'advance' version of the code (basically, the
> 'stand-alone test program' I have used so far) to the list in order to
> enable some third-party review. The code is, however, presently
> completely comment-free (could be added to some reasonable degree, eg
> no 'fuck me gently with a chainsaw'-style witticisms) and presumably
> somewhat dense to people who are not used to regard 'integers' a
> finite, ordered bit-sets.

Hi Rainer,

please do (show an advance/preview version) - if you don't want to expose  
it to the list yet for whatever reason, but want my comments, feel free to  
send personal mail.

WRT commentary, I think I am with Stroustrup here, meaning that the  
commentary is good to fill in what *cannot* be better said in code.  
Instead, comments should provide a general idea of what a piece of code  
does, and possibly describe the interfaces and return values, at least the  
special cases (if any) such as NULL, -1, 0, ...  descriptive and accurate  
variable and function names also reduce the need for commentary in epic  
breadth. :-)

Best regards

-- 
Matthias Andree