diff options
| author | Tom Lane <tgl@sss.pgh.pa.us> | 2005-12-22 22:50:14 +0000 |
|---|---|---|
| committer | Tom Lane <tgl@sss.pgh.pa.us> | 2005-12-22 22:50:14 +0000 |
| commit | faefc812f8aa41c4875520640db0c4e6d5c2edca (patch) | |
| tree | 6378c5fc579431547d0d8574e66b4b97beb2e8de /src/backend/utils/adt/varchar.c | |
| parent | ba6e1b2e729760f8a8862e7301350f6776d83ae5 (diff) | |
Adjust string comparison so that only bitwise-equal strings are considered
equal: if strcoll claims two strings are equal, check it with strcmp, and
sort according to strcmp if not identical. This fixes inconsistent
behavior under glibc's hu_HU locale, and probably under some other locales
as well. Also, take advantage of the now-well-defined behavior to speed up
texteq, textne, bpchareq, bpcharne: they may as well just do a bitwise
comparison and not bother with strcoll at all.
NOTE: affected databases may need to REINDEX indexes on text columns to be
sure they are self-consistent.
Diffstat (limited to 'src/backend/utils/adt/varchar.c')
| -rw-r--r-- | src/backend/utils/adt/varchar.c | 20 |
1 files changed, 14 insertions, 6 deletions
diff --git a/src/backend/utils/adt/varchar.c b/src/backend/utils/adt/varchar.c index eeb218bf57a..40e771ffab5 100644 --- a/src/backend/utils/adt/varchar.c +++ b/src/backend/utils/adt/varchar.c @@ -8,7 +8,7 @@ * * * IDENTIFICATION - * $PostgreSQL: pgsql/src/backend/utils/adt/varchar.c,v 1.108 2004/12/31 22:01:22 pgsql Exp $ + * $PostgreSQL: pgsql/src/backend/utils/adt/varchar.c,v 1.108.4.1 2005/12/22 22:50:14 tgl Exp $ * *------------------------------------------------------------------------- */ @@ -551,11 +551,14 @@ bpchareq(PG_FUNCTION_ARGS) len1 = bcTruelen(arg1); len2 = bcTruelen(arg2); - /* fast path for different-length inputs */ + /* + * Since we only care about equality or not-equality, we can avoid all + * the expense of strcoll() here, and just do bitwise comparison. + */ if (len1 != len2) result = false; else - result = (varstr_cmp(VARDATA(arg1), len1, VARDATA(arg2), len2) == 0); + result = (strncmp(VARDATA(arg1), VARDATA(arg2), len1) == 0); PG_FREE_IF_COPY(arg1, 0); PG_FREE_IF_COPY(arg2, 1); @@ -575,11 +578,14 @@ bpcharne(PG_FUNCTION_ARGS) len1 = bcTruelen(arg1); len2 = bcTruelen(arg2); - /* fast path for different-length inputs */ + /* + * Since we only care about equality or not-equality, we can avoid all + * the expense of strcoll() here, and just do bitwise comparison. + */ if (len1 != len2) result = true; else - result = (varstr_cmp(VARDATA(arg1), len1, VARDATA(arg2), len2) != 0); + result = (strncmp(VARDATA(arg1), VARDATA(arg2), len1) != 0); PG_FREE_IF_COPY(arg1, 0); PG_FREE_IF_COPY(arg2, 1); @@ -692,7 +698,9 @@ bpcharcmp(PG_FUNCTION_ARGS) * bpchar needs a specialized hash function because we want to ignore * trailing blanks in comparisons. * - * XXX is there any need for locale-specific behavior here? + * Note: currently there is no need for locale-specific behavior here, + * but if we ever change the semantics of bpchar comparison to trust + * strcoll() completely, we'd need to do something different in non-C locales. */ Datum hashbpchar(PG_FUNCTION_ARGS) |