> On Sun, 2008-09-21 at 22:26 +0200, Kinkie wrote: > >> Still, WordList is ready for merging, while String a long way off. >> It's build-tested, it's run-tested, and it's a fairly straightforward >> porting of the current codebase. >> It may be a waste of time in the end, but if it's mergeable, not >> merging it makes it doubly so. > > WordList class, in its current shape, is not something I would like to > see used in Squid, primarily because it violates several basic > requirements for a good C++ class. It even violates the only two > mandatory requirements listed on the Squid3CodingGuidelines wiki page! > WordList widespread use in the code is an aggravating factor in this > case. > > If others insist that this work must continue now, I may resume the > review to eliminate the bugs I see. I would very much prefer not to be > forced to spend time on that because I doubt we need that custom class > at all. >
I'm fence-sitting on this one. I never saw the need beyond perfomance, and String gains blow that well out of the water. Amos >> Okay, so what next? >> >> Should I fix the major issues Alex has mentioned in WordList >> (documentation and moving out of algorithms/) and merge it, or abandon >> it? > > I do not know what you mean by merging, but my recommendation is to > suspend your WordList project until we have a better class for a word > than "char*". At that time, I would return to the question whether we > can just use std::list or similar. > > HTH, > > Alex. > > >
