Author: overlordq
Description:
The SQL starting on line 162:
$sql2 = "SELECT * FROM $recentchanges $forceclause".
($uid ? "LEFT OUTER JOIN $watchlist ON wl_user={$uid} AND wl_title=rc_title AND wl_namespace=rc_namespace " : "") .
"WHERE rc_timestamp >= '{$cutoff}' {$hidem} " .
"ORDER BY rc_timestamp DESC";Does not work in current versions of PostgreSQL and will not work in future versions.
Version: 1.12.x
Severity: normal
| Status | Subtype | Assigned | Task | |
|---|---|---|---|---|
| · · · | ||||
| Invalid | None | T2384 [REPLACED BY TAG] PostgreSQL/pgsql support (tracking) | ||
| Resolved | None | T14784 Implicit Casts Removed in Postgresql 8.3 | ||
| Resolved | None | T14910 MySQL Specific SQL in SpecialRecentchanges | ||
| · · · |
overlordq wrote:
This is due to the fact that in the recentchanges table rc_bot is a character(1) field, and you're trying to compare it to a 1 or a 0.
This can either be rectified by
A) Using an actual integer datatype for the table in PG
-or-
B) Using a cast
Without digging deeper, option A would require no code changes here, but might elsewhere, while option B would require changing code.
You're using a 1-character text string for boolean fields? Ok, you guys can stop making fun of us for our funny timestamp format now. ;)
overlordq wrote:
I believe this was fixed in r30800 and Greg just forgot to close this bug.
(In reply to comment #2)
You're using a 1-character text string for boolean fields? Ok, you guys can
stop making fun of us for our funny timestamp format now. ;)
We only did that because you guys don't have a real boolean field, and assign 1 and 0 to your pseudo-booleans in the code. :) No worries, PG is using smallint everywhere mysql is using bool/tinyint now, so everyone should be happy.