Hi Paul, Paul Eggert wrote: > Attached are two proposed performance tweaks I found by inspection. No > big deal of course.
Thanks. I committed the first one in your name, with a reference to the precise ISO C section. Then I found that POSIX's "Portable character set" goes beyond that, and thus applied the patch below. Your second patch is then not needed any more. https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/basedefs/V1_chap06.html section 6.1 says "The encoded values associated with the members of the portable character set are each represented in a single byte.", and this "portable character set goes beyond what ISO C has. 2023-07-04 Bruno Haible <br...@clisp.org> mbchar: Optimize all chars from the POSIX "portable character set". * lib/mbchar.h (is_basic): Include all the characters from the POSIX "portable character set". * lib/mbchar.c (is_basic_table): Likewise. * lib/mbiter.h (mbiter_multi_next): Update comment. * lib/mbuiter.h (mbuiter_multi_next): Likewise. * lib/mbfile.h (mbfile_multi_getc): Likewise. diff --git a/lib/mbchar.c b/lib/mbchar.c index 84b388bfee..63ff9c7a72 100644 --- a/lib/mbchar.c +++ b/lib/mbchar.c @@ -24,13 +24,12 @@ #if IS_BASIC_ASCII -/* Bit table of characters in the ISO C "basic character set", - plus the characters '@', '$', and '`' which - ISO C guarantees to be single-byte and in practice are safe - to treat as basic in the execution character set. */ +/* Bit table of characters in the POSIX "portable character set", which + POSIX guarantees to be single-byte and in practice are safe to treat + like the ISO C "basic character set". */ const unsigned int is_basic_table [UCHAR_MAX / 32 + 1] = { - 0x00001a00, /* '\t' '\v' '\f' */ + 0x00003f81, /* '\0' '\007' '\010' '\t' '\n' '\v' '\f' '\r' */ 0xffffffff, /* ' '......'?' */ 0xffffffff, /* '@' 'A'...'Z' '[' '\\' ']' '^' '_' */ 0x7fffffff /* '`' 'a'...'z' '{' '|' '}' '~' */ diff --git a/lib/mbchar.h b/lib/mbchar.h index dea1e462ee..36bae18276 100644 --- a/lib/mbchar.h +++ b/lib/mbchar.h @@ -272,35 +272,40 @@ mb_copy (mbchar_t *new_mbc, const mbchar_t *old_mbc) } -/* is_basic(c) tests whether the single-byte character c is in the - ISO C "basic character set" or is one of '@', '$', and '`' which - ISO C 23 § 5.2.1.1.(1) guarantees to be single-byte and in practice - are safe to treat as basic in the execution character set. +/* is_basic(c) tests whether the single-byte character c is + - in the ISO C "basic character set" or is one of '@', '$', and '`' + which ISO C 23 § 5.2.1.1.(1) guarantees to be single-byte and in + practice are safe to treat as basic in the execution character set, + or + - in the POSIX "portable character set", which + <https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap06.html> + equally guarantees to be single-byte. This is a convenience function, and is in this file only to share code - between mbiter_multi.h and mbfile_multi.h. */ -#if (' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) && ('$' == 36) \ - && ('%' == 37) && ('&' == 38) && ('\'' == 39) && ('(' == 40) \ - && (')' == 41) && ('*' == 42) && ('+' == 43) && (',' == 44) \ - && ('-' == 45) && ('.' == 46) && ('/' == 47) && ('0' == 48) \ - && ('1' == 49) && ('2' == 50) && ('3' == 51) && ('4' == 52) \ - && ('5' == 53) && ('6' == 54) && ('7' == 55) && ('8' == 56) \ - && ('9' == 57) && (':' == 58) && (';' == 59) && ('<' == 60) \ - && ('=' == 61) && ('>' == 62) && ('?' == 63) && ('@' == 64) && ('A' == 65) \ - && ('B' == 66) && ('C' == 67) && ('D' == 68) && ('E' == 69) \ - && ('F' == 70) && ('G' == 71) && ('H' == 72) && ('I' == 73) \ - && ('J' == 74) && ('K' == 75) && ('L' == 76) && ('M' == 77) \ - && ('N' == 78) && ('O' == 79) && ('P' == 80) && ('Q' == 81) \ - && ('R' == 82) && ('S' == 83) && ('T' == 84) && ('U' == 85) \ - && ('V' == 86) && ('W' == 87) && ('X' == 88) && ('Y' == 89) \ - && ('Z' == 90) && ('[' == 91) && ('\\' == 92) && (']' == 93) \ - && ('^' == 94) && ('_' == 95) && ('`' == 96) && ('a' == 97) && ('b' == 98) \ - && ('c' == 99) && ('d' == 100) && ('e' == 101) && ('f' == 102) \ - && ('g' == 103) && ('h' == 104) && ('i' == 105) && ('j' == 106) \ - && ('k' == 107) && ('l' == 108) && ('m' == 109) && ('n' == 110) \ - && ('o' == 111) && ('p' == 112) && ('q' == 113) && ('r' == 114) \ - && ('s' == 115) && ('t' == 116) && ('u' == 117) && ('v' == 118) \ - && ('w' == 119) && ('x' == 120) && ('y' == 121) && ('z' == 122) \ - && ('{' == 123) && ('|' == 124) && ('}' == 125) && ('~' == 126) + between mbiter.h, mbuiter.h, and mbfile.h. */ +#if (' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \ + && ('$' == 36) && ('%' == 37) && ('&' == 38) && ('\'' == 39) \ + && ('(' == 40) && (')' == 41) && ('*' == 42) && ('+' == 43) \ + && (',' == 44) && ('-' == 45) && ('.' == 46) && ('/' == 47) \ + && ('0' == 48) && ('1' == 49) && ('2' == 50) && ('3' == 51) \ + && ('4' == 52) && ('5' == 53) && ('6' == 54) && ('7' == 55) \ + && ('8' == 56) && ('9' == 57) && (':' == 58) && (';' == 59) \ + && ('<' == 60) && ('=' == 61) && ('>' == 62) && ('?' == 63) \ + && ('@' == 64) && ('A' == 65) && ('B' == 66) && ('C' == 67) \ + && ('D' == 68) && ('E' == 69) && ('F' == 70) && ('G' == 71) \ + && ('H' == 72) && ('I' == 73) && ('J' == 74) && ('K' == 75) \ + && ('L' == 76) && ('M' == 77) && ('N' == 78) && ('O' == 79) \ + && ('P' == 80) && ('Q' == 81) && ('R' == 82) && ('S' == 83) \ + && ('T' == 84) && ('U' == 85) && ('V' == 86) && ('W' == 87) \ + && ('X' == 88) && ('Y' == 89) && ('Z' == 90) && ('[' == 91) \ + && ('\\' == 92) && (']' == 93) && ('^' == 94) && ('_' == 95) \ + && ('`' == 96) && ('a' == 97) && ('b' == 98) && ('c' == 99) \ + && ('d' == 100) && ('e' == 101) && ('f' == 102) && ('g' == 103) \ + && ('h' == 104) && ('i' == 105) && ('j' == 106) && ('k' == 107) \ + && ('l' == 108) && ('m' == 109) && ('n' == 110) && ('o' == 111) \ + && ('p' == 112) && ('q' == 113) && ('r' == 114) && ('s' == 115) \ + && ('t' == 116) && ('u' == 117) && ('v' == 118) && ('w' == 119) \ + && ('x' == 120) && ('y' == 121) && ('z' == 122) && ('{' == 123) \ + && ('|' == 124) && ('}' == 125) && ('~' == 126) /* The character set is ISO-646, not EBCDIC. */ # define IS_BASIC_ASCII 1 @@ -320,7 +325,9 @@ is_basic (char c) { switch (c) { - case '\t': case '\v': case '\f': + case '\0': + case '\007': case '\010': + case '\t': case '\n': case '\v': case '\f': case '\r': case ' ': case '!': case '"': case '#': case '$': case '%': case '&': case '\'': case '(': case ')': case '*': case '+': case ',': case '-': case '.': case '/': diff --git a/lib/mbfile.h b/lib/mbfile.h index 716ab3fc89..9a2532992e 100644 --- a/lib/mbfile.h +++ b/lib/mbfile.h @@ -108,9 +108,12 @@ mbfile_multi_getc (struct mbchar *mbc, struct mbfile_multi *mbf) /* Handle most ASCII characters quickly, without calling mbrtowc(). */ if (mbf->bufcount == 1 && mbsinit (&mbf->state) && is_basic (mbf->buf[0])) { - /* These characters are part of the basic character set. ISO C 99 - guarantees that their wide character code is identical to their - char code. The 32-bit wide character code is the same as well. */ + /* These characters are part of the POSIX portable character set. + For most of them, namely those in the ISO C basic character set, + ISO C 99 guarantees that their wide character code is identical to + their char code. For the few other ones, this is the case as well, + in all locale encodings that are in use. The 32-bit wide character + code is the same as well. */ mbc->wc = mbc->buf[0] = mbf->buf[0]; mbc->wc_valid = true; mbc->ptr = &mbc->buf[0]; diff --git a/lib/mbiter.h b/lib/mbiter.h index fadefe104b..bc88b4f3a2 100644 --- a/lib/mbiter.h +++ b/lib/mbiter.h @@ -124,9 +124,12 @@ mbiter_multi_next (struct mbiter_multi *iter) /* Handle most ASCII characters quickly, without calling mbrtowc(). */ if (is_basic (*iter->cur.ptr)) { - /* These characters are part of the basic character set. ISO C 99 - guarantees that their wide character code is identical to their - char code. */ + /* These characters are part of the POSIX portable character set. + For most of them, namely those in the ISO C basic character set, + ISO C 99 guarantees that their wide character code is identical to + their char code. For the few other ones, this is the case as well, + in all locale encodings that are in use. The 32-bit wide character + code is the same as well. */ iter->cur.bytes = 1; iter->cur.wc = *iter->cur.ptr; iter->cur.wc_valid = true; diff --git a/lib/mbuiter.h b/lib/mbuiter.h index 954e11f635..93dec81603 100644 --- a/lib/mbuiter.h +++ b/lib/mbuiter.h @@ -132,9 +132,12 @@ mbuiter_multi_next (struct mbuiter_multi *iter) /* Handle most ASCII characters quickly, without calling mbrtowc(). */ if (is_basic (*iter->cur.ptr)) { - /* These characters are part of the basic character set. ISO C 99 - guarantees that their wide character code is identical to their - char code. */ + /* These characters are part of the POSIX portable character set. + For most of them, namely those in the ISO C basic character set, + ISO C 99 guarantees that their wide character code is identical to + their char code. For the few other ones, this is the case as well, + in all locale encodings that are in use. The 32-bit wide character + code is the same as well. */ iter->cur.bytes = 1; iter->cur.wc = *iter->cur.ptr; iter->cur.wc_valid = true;
