| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Comments:
<0> Darien: that's why I don't bother cleaning up after myself.
<0> I'm just going to make a mess again later.
<1> exactly
<1> just re-use the same mess
<0> Darien: actually, I think that IS what apr does with memory pools and threading...
<1> 'I'm takin' / Birthday cake 'n' / Chili and greasy old bacon / Throw it all on top of the mess I been making / Wife's so mad she starts to shakin'
<0> though being a c++ kinda guy I prefer new and delete over malloc()
<2> I cant belive that mysql is that unpredictable.
<1> MYSQL isn't unpredictable
<3> Snadder: Eh?
<0> Snadder: I can't believe that computers continue to surprise you.
<0> to err is human, to really **** things up takes a computer
<1> garbage in, garbage out
<0> and of course, to err is human, to moo, bovine.
<4> with incredible speed and accuracy
<2> is the change from null to '' on a update apart of the sql standar?
<0> memset(random_address,random_data,random_size);!
<3> Snadder: No.
<1> Snadder: no
<2> whats the point doing that then?
<1> in order to not reject queries
<2> wont it only confuse people like me
<1> but you can change the SQL mode if that's not what you want
<0> Snadder: I guess you just answered your own question
<3> Snadder: If you're a MySQL newbie, yes.
<0> ...
<2> [RainMkr], it will not only confuse newbie's.. but it will also confuse people used to other database engines.. like postgresql (like myself)
<3> Snadder: please note my statement "... MySQL newbie ..."
<2> hehe. ok.
<0> Snadder: anyway, I thought anybody who used postgres was so smug they enjoyed the smell of their own farts. *grin*
<3> Snadder: You are obviously not a DB newbie. If you where, you wouldn't be bitching about this behavior. :)
<0> but I never disliked a pgsql user unless he was george clooney
<5> they killed chef!
<5> you bastards.
<0> thank you
<2> [RainMkr], exactly ;-)
<0> somebody got it ;-)
<5> i go eat dinner now!
<6> Old news ..
<0> no news
<0> is good news?
<2> is there any other settings I can set.. to get mysql to operate more like postgresql?.. or more "strict"?
<1> LISTEN TO WHAT I AM TELLING YOU
<1> CHANGE THE SQL SERVER MODE
<2> yes.. I did
<1> !m Snadder sql mode
<7> Snadder: (Selecting SQL Modes) : http://dev.mysql.com/doc/mysql/en/SQL_mode.html
<1> this is it
<1> this right here
<1> this is what you can do
<2> yes.. I already did: SET SESSION sql_mode='TRADITIONAL'
<2> but I wonder if there is any other options other than this
<1> you could try setting it to 'POSTGRESQL'
<1> there are TONS
<1> if you'd read the page that I gave you, you'd know that
<2> other than the session variable?
<2> i've read it..
<1> you mean like setting it globally?
<2> nopes.. anything else which might get the db to be more like postgresql. other than the global/session variable
<2> but I guess there is no more.. after what your telling me.
<2> Darien, I had to use TRADITIONAL cause POSTGRESQL didnt include STRICT_TRANS_TABLES or STRICT_ALL_TABLES
<2> which solves my problem.
<8> anyone know if there are any tools to find out theh performance of a 'named' keycache? it seems the 'show status like '%key%' doesn't list them
<9> I need to have mysql search through large bodies of text efficiently, anyone know of articles on how to do that?
<10> !man fulltext
<7> (Full-Text Search Functions) : http://dev.mysql.com/doc/mysql/en/Fulltext_Search.html
<11> wheee!... I just made my first function :)
<11> mysql function that is.
<11> ip_to_hex to store IPs in binary() fields using 8 bytes.
<11> and hex_to_ip ofcourse to read it out again.
<4> stored func or udf?
<11> stored function.
<11> not sure how to do a udf yet. I
<11> 'll have to have a friend of mine who is better at C than me do it for me.
<4> I started doing a udf but I got distracted
<4> seems simple enough
<11> PostgreSQL has an IP/Network field data type.
<11> I was trying to symulate that functionality (to a lesser degree).
<4> I got as far as writing a stub that would compile
<11> I'm sure their way will only use 4 bytes for the IP and 4 bytes for the netmask. The way I have it will cost 8 bytes for IP and 8 for netmask
<9> mhillyer: how do I run more precise searches than fulltext search functions give me? Like I want to be able to match "this query" but only if it is also near "that query" ?
<10> It is a tradeoff, you wither want accuracy or speed. Fulltext is speed.
<8> near 'that query?'?
<9> mhillyer: right but it is a natural language search right?
<10> Yes it is.
<8> err.. near*
<9> mhillyer: there is no optimized exact language search?
<8> what do you mean by 'near'
<9> zircu: like same sentence
<10> reflous: WHERE textcol LIKE 'this string';
<10> That is exact.
<9> mhillyer: but slow right?
<8> reflous: are you doing a BOOLEAN match?
<10> If you actually want an exact string match and index for it, it is not slow.
<9> mhillyer: hmm ok
<8> reflous: no that will use an index, unless it is like '%this string'
<9> ah, yeah I'd want to be able to use things like: incorpat% so I get incorporation and incorporated
<10> reflous: As long as the wildcard is at the end of the string.
<9> erh s/incorpat%/incorporat%/ :P
<2> is it then possible to find out which text is nearest the keyword?
<9> ok so ... WHERE col1 like 'incorporat%' AND MATCH (col2) AGAINST ('bylaws'); would match sentence: "every corporation should have good bylaws"
<2> like.. if searched with inc% I want to display inc higher than incorporation
<8> not with a 'like' pattern
<2> ..as an example.
<10> Whoa, don't go mixing searches.
<9> inefficient?
<10> And the match you just gave as an example does not start with 'incorporat
<9> ah yeah, I'm tired
<8> the fulltext search will sort by a relevance of the match you you include it in your where or order by
<2> zircu, how is the fulltext search syntax?
<9> zircu: yeah, but I need the ability to insert exact matches... like the user needs to be abl eto say: "this phrase" plus no phrase but these words
<8> Snadder: how do you mean?
<8> reflous: well if you p*** '"this phrase'" to the against it will find that exact phrase
<8> if you want incorporat to match differnt things you can do AGAINST 'incorporat*'
<2> zircu, ok.. I found the docs about it.. thanks.
<12> Hi! Question for the pros ... (:
<9> zircu: i get strange results with fulltext searches, the boolean mode looks sweet, can't understand these results though
<12> I created a trigger, `after delete', to removed ***ociated records from another table.
<12> It tells me "table x was not locked with LOCK TABLES"
<8> reflous: the boolean doesn't have a relevance of match, it returns what ever matched
<9> zircu: I'm actually getting less matches then I should
<8> reflous: boolean is more strict
<9> zircu: ah
<12> do I really need to lock the entire table?
<8> lclark_: you have a LOCK table statement but it doesn't include the x table that you used till you issued the unlock
<8> or it doesn't match an alias you use
<13> i have to find a range between 2 dates. the dates are strings (mm/dd/yy) is there a way to use str_to_date to get my result or should I figure out how to change that column into a datetime (using mysql5)
<9> zircu: no that doesn't explain it either... I entered the fulltext example that is given on the mysql webpage, so I have a row: '1. Never run mysqld as root. 2. ...' i then run: SELECT * FROM articles WHERE MATCH (title,body) AGAINST ('Never' IN BOOLEAN MODE); and i get 0 results
<8> lclark_: iirc you might want a innoDB type db to do row level locking
<12> zircu: conveniently enough, I've used innodb for the FK fanciness.
<8> i havn't used innodb much to know details on the problem
<8> reflous: hmm.. i wonder if never is a stop word
<9> zircu: oh weird, maybe :P
<8> oh yea it is
<8> http://dev.mysql.com/doc/refman/4.1/en/fulltext-stopwords.html
<9> zircu: what does it mean to be a "stop word"
<8> reflous: they are ignored, basicall if you type 'Never mysqld' mysql just will use 'mysqld' to match with
<8> you can customize that list with a text file and a param in your my.cnf file
<9> zircu: ahh ok good
<8> i belive the index will need to get rebuilt after you do that as well
<9> zircu: is there a way to get google type results, where you might have a paragraph, but the search only returns that relevant sentence? or would you need to do that with a combo of a select fulltext search with a script parsing down the result to the sentence?
<6> the later
<9> hmmm, that could be pretty inefficient
<8> reflous: yeah you have to do some parsing yourself
<9> ok
<6> reflous: that's why google don't use mysql for their search results :)
<9> hmmm, they use their own db right?
<8> i have a fulltext search that will take the words like 'mysql server' and then convert it to 'mysql <mysql* server <server*' so it will bring back both a match of 'mysql server' and 'mysql servers' but the first one will get listed first
<8> reflous: they have a rather complicated system
<8> i'm always amazed how they can get a search result of 'about 2 million' records in 0.002 seconds
Return to
#mysql
or
Go to some related
logs:
rsynth gentoo say
conky ion3
python
#osdev
#mysql
gentoo moznoxft firefox
mysql string_to_date
#web
#perl
#ubuntu
|
|