Author: Dave Gomboc
Date: 01:08:47 11/29/02
Go up one level in this thread
On November 28, 2002 at 23:41:52, Uri Blass wrote: >On November 28, 2002 at 21:57:59, Dave Gomboc wrote: > >>On November 28, 2002 at 19:23:36, David Rasmussen wrote: >> >>>On November 28, 2002 at 17:36:22, Dave Gomboc wrote: >>> >>>> >>>> >>>>I posted a corrected version of my post where I included a half-smiley. This is >>>>because the comment is half a joke, but also half serious. >>>> >>>>Philisophically, the main search already needs to deal with how best to search >>>>at huge depths vs. how best to search at tiny depths. Some search code ignores >>>>the difference; it is these cases that probably depend on the presence of a >>>>separate q-search the most. Search code which is designed to be adaptive >>>>according to search depth should not have trouble encompassing q-search as well. >>>> >>>>Practically, it may still be clearer to express what needs to be done near the >>>>tips with a specific q-search routine. >>>> >>> >>>That was sort of my points also: In principle, there aren't any differences >>>between normal search and qsearch. Qsearch can be "expressed" within the >>>framework of normal search with pruning/extensions. My other point was, that >>>design matters a lot, and that what might be "semantically" equivalent, might >>>not be when it comes to implementing, and/or expressing what to be done. A >>>separate qsearch function is a very good idea, designwise, IMO. >>> >>>/David >>Hmm. My gut feeling is that a well-written main search makes a distinct >>q-search routine of no benefit, and that separate q-search routines are harmful >>to program development because they allow the developer to be lazy when >>implementing the main search. >> >>Dave > >I do not understand > >If you want different rules at different plies you can have many search >functions(search at remaining depth 1,search at remaining depth 2,....) and you >can have one function. > >It is only a question of style of writing. > >The developer is always allowed to be lazy in implementing the main search even >without qsearch. > >I do not see how not having qsearch help the programmer not to be lazy. > >Uri It's not a matter of how you divvy up the code, it's a matter of how you think about your code. Dave
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.