Closed mcostalba closed 5 years ago
In position.h, from_to_bb should be fromToBB.
• In has_game_cycle(), to and from should be s1 and s2 • trailing spaces in line 1003 of search.cpp
In the comment here: https://github.com/official-stockfish/Stockfish/commit/3d6995eae8039b2bf4141cbc02d87d5a6c2a1905
we need a space between "//" and "Increase"
In pawns.cpp, the three basic pawn penalties (Isolated, Backward, and Doubled) area clearly self documenting and don't need their own special comments/spaces. . I propose:
//Pawn Penalties constexpr Score Isolated = S(13,16); constexpr Score Backward = S(17, 11); constexpr Score Doubled = S(13,40);
Tabs and spaces in line 1167 of search.cpp
line 779 of evaluate.cpp, there is an extra space -136 ; --> -136;
https://github.com/official-stockfish/Stockfish/blob/a834bfe8332adcb0dfc1fd280f1f9d8bbce86266/src/search.cpp#L953-L955 Since PawnValueEg * (depth / ONE_PLY) is of type Value, the cast to Value is superfluous and should be removed.
In pawns.cpp, lines 224 and 227, the default value for ourRank and theirRank should probably be RANK_1 instead of 0.
@protonspring In that case, it seems that ourRank and theirRank should also be of type Rank instead of int.
I agree with that.
On Sat, Jun 23, 2018 at 4:31 PM, syzygy1 notifications@github.com wrote:
@protonspring https://github.com/protonspring In that case, it seems that ourRank and theirRank should also be of type Rank instead of int.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/official-stockfish/Stockfish/issues/1426#issuecomment-399714775, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSNeFBvJGbS3kCxcGbQ44SGV3FWu0ks5t_sHegaJpZM4SR1Zw .
-- Michael T. Whiteley mobile: 801.707.6886
in evaluate.cpp // Bonus for king attacks on pawns or pieces which are not pawn-defended
should be something like // Bonus for king attacks on pawns or pieces which are not strongly protected
or more simply, remove this comment.
in evaluate.cpp the Overload bonus and the following comment // Bonus for overload (non-pawn enemies attacked once or more and defended exactly once)
Technically, a piece is "overloaded" (or "overworked") if it has more than one defensive duty. For example, it is needed to defend 2 or more squares, and moving to one such square would leave the other poorly defended. It might also be for example a rook which is needed on the back rank otherwise a mate will follow, and thus cannot defend squares along the file.
This is not exactly what we are computing here.
Suggestions? I was thinking of Pressured, or Tied
Drop
from evaluate.h
In position.h line 479
if (!can_castle(WHITE) && !can_castle(BLACK))
can be replaced with
if (!can_castle(ANY_CASTLING))
as we do in search.cpp line 650 and 1670
the name "can_castle", although short and sweet, is misleading in 2 ways a) it returns an int, not a bool b) it is only about remaining castling rights, not about the legality of a next move castling
Maybe the longer "castling_rights" would leave no surprise.
Also note that int can_castle(CastlingRight cr) const; is always used as a bool
int can_castle(Color c) const; would always be used as a CastlingRight (an accessor to record the current castling right for a given color) after the change proposed in the post just above
In misc.cpp line 231 int get_group(size_t idx) {
this was taken from Texel code
However, in SF, nowhere else do we use names such as get or find
Maybe something like bestGroupForThread
Line 577 of evaluate.cpp says: "Keep only the squares which are not completely unsafe." This could be much more clear with something more like: "Keep only the squares which are relatively safe." or similar.
• pawns.cpp, lines 35-37: fix spaces, use alphabetical order • evaluate.cpp: better comments for king tropism
material.cpp line 62 there is an extra space before ==
&& pos.count<QUEEN>(us) == 1
In the README, the links to the chess programming wiki should be updated to https://www.chessprogramming.org/Main_Page and https://www.chessprogramming.org/Stockfish
between_bb() is still used, so BetweenBB[][] too. Or am I missing something?
You are right. I will delete my comment !
In evaluate .cpp, we can replace
nonPawnEnemies = pos.pieces(Them) ^ pos.pieces(Them, PAWN);
with
nonPawnEnemies = pos.pieces(Them) & ~pos.pieces(PAWN);
Because under the hood we have this for the first
byColorBB[c] ^ (byColorBB[c] & byTypeBB[pt])
And this for the proposed change, so less instructions for the same result
byColorBB[c] & ~byTypeBB[pt]
from PR #1774
In search.cpp
:
// Singular extension search (~60 Elo). If all moves but one fail low on a
// search of (alpha-s, beta-s), and just one fails high on (alpha, beta),
// then that move is singular and should be extended. To verify this we do
// a reduced search on on all the other moves but the ttMove and if the
// result is lower than ttValue minus a margin then we will extend the ttMove.
line 913:
&& !excludedMove // Recursive singular search is not allowed
The comment is misleading, since recursive singular search is allowed and wanted by design, we just not want a vicious cycle at the same ply, so maybe better:
&& !excludedMove // needed to avoid a vicious circle
The term rBeta is used twice in search:
This confuses me every time I glance at this variables and prevents me to memorize this 2 different techniques correctly (do we raise or reduce beta in probcut? and what on singular search? ). Using explicit variable names here help's IMO to keep things clear in mind: Probcut -> raisedBeta Singular Extension -> reducedBeta
There is a c missing in 'aquire', and I think it would be easier to read if "to avoid a thread reads" is changed to "to avoid that a thread reads".
Oh, and in line 1113 exsist should be exist.
There are some TABS in search.cpp that should probably be converted to spaces.
lines 451, 1189, 1190, and 1191.
Time for some roll up. Is there someone interested in flushing this log?
After a possible cumulative patch is applied I will open a new thread. Eventually to avoid threads to become too long and to allow easier understanding of current status, we should close the old one and reopen a new one every time the corresponding cumulative formatting patch is applied.
I can. Should i make a patch that incorporates all of them?
On Sun, Dec 9, 2018, 1:21 AM Marco Costalba <notifications@github.com wrote:
Time for some roll up. Is there someone interested in flushing this log?
After a possible cumulative patch is applied I will open a new thread. Eventually to avoid threads to become too long and to allow easier understanding of current status, we should close the old one and reopen a new one every time the corresponding cumulative formatting patch is applied.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/official-stockfish/Stockfish/issues/1426#issuecomment-445519400, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSNiZ9e-HH5wTZObSHGSvQw_NVTRxks5u3MgUgaJpZM4SR1Zw .
I made a PR with all of the changes I could find in this issue.
https://github.com/official-stockfish/Stockfish/pull/1861
If you have additions, please either post them in the PR so they can be added, or wait for another (clean) RENAMING issue to be added.
@protonspring great! I am going to close this one and reopen a new one
Use this issue to post renaming suggestions, once in a while, if approved will be committed at once.
We will use a table to make it easier account pending renames.
Please add your renames as new rows in the table. Thanks.