From: <And...@wi...> - 2003-01-30 20:49:52
|
Hi. After compiling the lastest (01/26) snapshot I was trying some searches and found that it would not find on double colons (as in a perl module use stmt) getopt::std retreived nothing, nor did: "getopt::std" while: getopt std was the same as: "getopt std" "getopt :: std" "getopt :: std ::" a Andy Bach, Sys. Mangler Internet: and...@wi... VOICE: (608) 261-5738 FAX 264-5030 " ... even if you're mediocre/decent at perl [the cmecf] code is pretty confusing in certain areas ..." CB |
From: Lachlan A. <lac...@ip...> - 2003-01-31 05:29:24
|
Greetings Andy, Thanks for your bug report. 1. What are your settings for valid_punctuation and =20 extra_word_characters? 2. Have you changed either since last rebuilding the database? (That=20 behaviour is expected if either contained ":" when the database was=20 built, but not when it is searched.) 3. Did it work for earlier versions? 4. Does it find "getoptstd"? 5. Have you tested it with single colons? Cheers, Lachlan On Friday 31 January 2003 07:49, And...@wi... wrote: > After compiling the lastest (01/26) snapshot I was trying some > searches and found that it would not find on double colons (as in a > perl module use stmt) > =09getopt::std > retreived nothing, > nor did: > =09"getopt::std" > while: > =09getopt std > was the same as: > =09"getopt std" > =09"getopt :: std" > =09"getopt :: std ::" |
From: Lachlan A. <lh...@us...> - 2003-01-31 05:32:47
|
On Friday 31 January 2003 16:25, Lachlan Andrew wrote: > 2. Have you changed either since last rebuilding the database?=20 > (That behaviour is expected if either contained ":" when the > database was built, but not when it is searched.) Errr... Make that 'if neither contained ":" when the database was=20 build, but valid_punctuation does when it is searched', or something=20 like that... L |
From: Lachlan A. <lh...@us...> - 2003-02-02 12:32:36
|
Greetings Andy/all, It seems this relates to an Undocumented Feature(tm). Words beginning=20 with 'exact:' or 'hidden:' are treated differently. As a side=20 effect, it introduces the bug of not splitting words at colons. Could someone who knows what exact: and hidden: mean please=20 explain what they are for (and/or document them officially)? I don't=20 want to break anything while trying to fix the bug. On a related note, does anyone have any ideas for the syntax of "field=20 restricted" searches? I was thinking of something like "title:word"=20 to search for "word" in the title field, or "heading:word" etc. Was=20 the plan to allow user-defined fields in meta-data to be searched? =20 That would be hard! Cheers, Lachlan On Friday 31 January 2003 07:49, And...@wi... wrote: > ...found that it would not find on double colons... > =09getopt::std > retreived nothing, as did: > =09"getopt::std" > while: > =09getopt std > was the same as: > =09"getopt std" > =09"getopt :: std" > =09"getopt :: std ::" |
From: Geoff H. <ghu...@ws...> - 2003-02-03 14:17:46
|
> Could someone who knows what exact: and hidden: mean please > explain what they are for (and/or document them officially)? I don't > want to break anything while trying to fix the bug. These are fuzzy algorithms essentially. You could have endings:blah. You're right that it's undocumented, and it should probably be taken out of the parser. (Nice idea to have per-word fuzzy possibilities, but maybe not the right way to do it.) > On a related note, does anyone have any ideas for the syntax of "field > restricted" searches? I was thinking of something like "title:word" > to search for "word" in the title field, or "heading:word" etc. Was > the plan to allow user-defined fields in meta-data to be searched? Well, this is the "normal" syntax used by other sites. As for user-defined fields, it's certainly an ultimate goal but I think it's more important to: a) get the new parser running. b) get typical field-restricted searches like "title:word" going. -Geoff |
From: Lachlan A. <lh...@us...> - 2003-02-06 09:52:56
|
Thanks for that, Geoff. Are these actually implemented? It seems to=20 me that exact: sets the isExact flag, which is never actually=20 read. (I was expecting it to disable the fuzzy algorithms, but it=20 doesn't, and grep doesn't show it being used anywhere.) Anyway, I'll leave that alone for now, and ask for feedback once the=20 patch is finished... Cheers, Lachlan On Tuesday 04 February 2003 01:17, Geoff Hutchison wrote: > > Could someone who knows what exact: and hidden: mean please > > explain what they are for? > > These are fuzzy algorithms essentially. You could have > endings:blah. You're right that it's undocumented, and it should > probably be taken out of the parser. (Nice idea to have per-word > fuzzy possibilities, but maybe not the right way to do it.) |
From: Lachlan A. <lh...@us...> - 2003-02-05 13:06:36
Attachments:
patch.colons
|
Greetings Andy, Try the attached patch. It is very rough-and-ready (since I'm in the=20 processes of also trying to add other functionality) but it should=20 get you going. Let me know how you get on. Cheers, Lachlan On Friday 31 January 2003 07:49, And...@wi... wrote: > ...it would not find on double colons... |