@# Quotes DB     useful, funny, interesting





Google
 
Web www.quotesdb.info
Undernet  |  EFnet  |  Quakenet  |  Freenode  |  Dalnet  |  Ircnet  |  Galaxynet
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20



Comments:

<0> yeah but an array is a single type
<0> limiting you somewhat
<1> things like Address1 address2, address3, could be in array I suppose
<2> can do multi-dimensional arrays tho
<2> to really freak people out :P
<0> yeah. I never heard of anyone that uses it :)
<1> o0o0o
<2> I have no use of it
<0> Mithandir: did you know FB can actually access blobs as real streams
<0> like they were files?
<2> but if I ever do, I'm going to be very thankful it's there, cause that means I'm doing something very weird :P
<3> ki9a: cool
<1> maybe useful to store two dimensional mappings
<0> just that almost nobody implemented the logic to do it
<0> Mithandir: they better just implement native guid support :)



<2> heheh
<2> well it's not hard to do with a UDF :P
<2> mind you, having a native GUID type would be cool
<0> yeah
<2> so you don't have to use the string representation
<0> and save more than half of the space it would take
<0> faster indexing wise
<0> storage wise
<2> save over 75% of the space actually
<0> you counted the {} 's? too
<4> so you guys are talking like a 8but unique identifier ?
<4> 8bit
<2> no, I stored as multibyte characters for some reason ... ignore me, it's late :P
<3> an 8 bit uid
<3> thats limited
<2> PoseFant: eh, GUID's are .. hmm 128 bit ?
<4> yes
<1> blade didn't detect the sarcasm :P
<4> and you want to make it smaller
<2> no
<4> so I suggested 8 bit
<2> currently to store a GUID you store the string representation as a VARCHAR
<2> which is 36 bytes
<3> why not just use quantum state storage?
<3> :P
<4> ah
<0> else it's 16
<4> I see
<3> i prefer as a string though
<0> why?
<1> hes english :P
<3> dunno really
<0> if it had native support for guid, all representations would look like strings
<3> force of habit
<1> he simple brain
<5> Ok ok ok... Why isn't my master/detail relation working...?
<4> it doesn't really matter tho
<3> he simple brain
<3> hah
<2> Mjolnir: it's signed the act against slavery
<3> your iq just overwhelms us
<1> laughs @ iq
<4> indexing varchars is not killing the system
<1> crum that up yours Blade^ :) you read too much :)
<2> PoseFant: indexes would get smaller and faster, considering GUID's are used mainly for linking, it adds quite a bit
<3> and you're just a ****
<3> weren't you banned?
<3> can we ban him again?
<3> i'll pay
<4> how would they get faster ?
<4> you change your queries because you change your guid ?
<0> smaller size of index element = faster
<3> TByte: of course it's tight, i don't get done up it each weekend unlike you ;p
<2> PoseFant: you'd on average have to compare half as many bits
<4> ki9a: not necessarily
<1> you finger yourself to test its tightness?
<3> yes
<3> its the only way
<4> Well, if you count CPU cycles used for a full index scan



<1> searching indexes is quite fast, isn't it...
<2> yes
<1> based on my RAM binary tree, it was stupidly fast :/
<1> I used int64s though :/
<5> Ah, doesn't work because my index field doesn't match my master field
<2> FB's generators are 64 biut
<0> PoseFant: you often have 2 or 3 guids in an index; especially in link tables.
<2> mind you, I usually use 32 bit fields for em cause I'm lazy
<1> I thnk my binary tree is the best thing I made, it's balanced and I had a lot of uses to it :/ inserting/deleting and searching in memory with tiny footprint
<0> I dont think I currently use generates.
<0> generators.
<5> 32 bit fields work fine especially if you use unique constraints.
<2> I use em quite a lot, in tables where briefcasing isn't needed
<4> ki9a: Ok thats fine, but it's linked tables
<2> PoseFant: yeah, liek I said, that's the main reason to use GUID's
<0> Mithandir: I only use guid these days.
<4> What really matters is what you want to select and where it is stored on disk
<4> you can have the supreme index, but you can kill the system if your data is spread around
<2> still beats having your data spread around without that index :)
<4> well
<4> consider you have a table of 200mill rows
<4> and you select 75million, spread around in that table
<1> heh I noticed when I pop a DB related question, people can't stop talking about dbs :/
<0> you are in trouble anyway
<4> do you think a full index scan and then a full table scan would be necessary ?
<0> if you want to select 75mln records at once.
<4> it would propably jump straight to the full table scan
<2> that depends on whether the index is in memory or not
<2> often they are
<4> what use does it have if it has to perform a full tablescan after the index scan ?
<2> it also depends on the selectiveness of the index etc
<2> why would it need to do a full table scan ?
<4> because its spread
<2> and ?
<4> it doesn't jump to a unique pointer on the disk with a unique RID in the index
<2> that's what the index is for, it tells the database where the data is spread
<4> it has to move through the table
<2> eh
<4> p*** the other rows
<2> taht depends entirely on the database
<4> if the data is spread and you select enough
<2> most store disk locations with their page info
<4> if you select say n..m
<4> an index would be great
<4> because it would then jump to n and select to m
<4> or even if you wanted to select n..m and k..l
<4> but when 90% of the rows are spread around
<4> index scanning is just waisting time
<4> may be
<1> I did this on a smaller scale :) TRBHardBinaryTree :) had data spread everywhere on the file :/
<4> as you said depends on the DBMS :P
<0> doesn't the record or page contain a "next page" ?
<1> indexed and everything :)
<2> y main job is data mining on a 30 Gb MS SQL database :P
<3> thats small
<1> ki9a yep it would :P
<3> unpaq the wikipedia db
<3> :P
<2> Blade^: it's probably less
<0> wikipedia isnt that large
<0> the images are 30 gb though
<0> or more
<2> our database is insanely bad
<3> its bzipped text
<0> they're not in the db
<3> of course it's larger
<0> Blade^: It's not 30 gb
<0> at least not the english one
<2> anyway
<2> off to do more packing
<0> Blade^: it's all .xml now
<6> http://oneredpaperclip.blogspot.com/
<0> it used to be sql
<6> now THAT is cool!


Name:

Comments:

Please enter the result of the sum 63 + 46 (to avoid spam):






Return to #delphi
or
Go to some related logs:

#cisco
upedia
#beginner
#beginner
team cheeto
#beginner
#politics
#photoshop
pass mycuteteens
#photoshop



Home  |  disclaimer  |  contact  |  submit quotes