|
infernal machines posted:are you suggesting lowtax wouldn't spring for a full stack team of 10x rockstars? lowatax will need them
|
# ? Mar 6, 2017 15:35 |
|
|
# ? May 15, 2024 04:41 |
|
come spring lowtax will need all our help working the bitfields
|
# ? Mar 6, 2017 18:01 |
|
George posted:come spring lowtax will need all our help working the bitfields
|
# ? Mar 6, 2017 18:15 |
|
George posted:come spring lowtax will need all our help working the bitfields
|
# ? Mar 6, 2017 18:26 |
|
I never understood what seems so clever to people about packing flags into bits in database integer columns. I've seen people at my last 4 jobs all do it and you could tell they were quite pleased with their cunning solution. one of them even gave a demo about how they used an integer to store a days of the month calendar you see because the longest month has 31 days and there are 32 bits so it's a perfect fit does any popular flavor of SQL even have bitwise operators
|
# ? Mar 6, 2017 18:40 |
|
AWWNAW posted:I never understood what seems so clever to people about packing flags into bits in database integer columns. I've seen people at my last 4 jobs all do it and you could tell they were quite pleased with their cunning solution. one of them even gave a demo about how they used an integer to store a days of the month calendar you see because the longest month has 31 days and there are 32 bits so it's a perfect fit uh
|
# ? Mar 6, 2017 18:43 |
|
this man was and is still the CTO
|
# ? Mar 6, 2017 18:46 |
|
Bitwise math is very cool and good when you're working in a resource constrained system like a microcontroller, but not being a database guy I don't know if it makes any significant difference on big servers
|
# ? Mar 6, 2017 18:59 |
|
Sagebrush posted:Bitwise math is very cool and good when you're working in a resource constrained system like a microcontroller, but not being a database guy I don't know if it makes any significant difference on big servers almost never. the cycles you waste doing the bitwise stuff is generally slower than just having boolean flags or whatever. the i/o isn't a problem because it will likely be fetched and cached at the same time if your architecture doesn't suck.
|
# ? Mar 6, 2017 19:36 |
|
i thought it was all about disk space, a boolean is a byte on disk under the best(worst?) case scenario bit packing will cut your storage by 87.5%
|
# ? Mar 6, 2017 19:49 |
|
Perplx posted:i thought it was all about disk space, a boolean is a byte on disk yeah, but think about the overall storage volume you're dedicating to settings vs data. if it's not a drop in the bucket there's something really wrong. 87.5% of .5% of your storage is nothing, so just throw a bit more disk at it.
|
# ? Mar 6, 2017 19:59 |
|
also that 1 byte per bit is if you only have 1 bit. sql server stores bits in 1 byte chunks so if you have 8 bit columns in a table that's still only 1 byte of storage
|
# ? Mar 6, 2017 20:02 |
|
25-32 bit fields are the same size as 1 int field.
|
# ? Mar 6, 2017 20:04 |
|
Shaggar posted:also that 1 byte per bit is if you only have 1 bit. sql server stores bits in 1 byte chunks so if you have 8 bit columns in a table that's still only 1 byte of storage nobody cares what sql server does because it's bad
|
# ? Mar 6, 2017 20:16 |
|
sql server is the best.
|
# ? Mar 6, 2017 20:16 |
|
lol. like I'm gonna use oracle. what am i, made of money??
|
# ? Mar 6, 2017 20:16 |
|
dude, have you heard of mongo?
|
# ? Mar 6, 2017 20:29 |
|
Perplx posted:i thought it was all about disk space, a boolean is a byte on disk What is the increase in time/money spent dealing with a code base littered with 'clever' snowflake crap like that that is somehow justified because in theory it could address a probably inconsequential future problem. In the case of disk space concerns which do you think would be the better approach? 1. Prematurely optimize your code in ways that make it difficult to understand, easy to break, and contribute to stagnating your application for over a decade despite at least hundreds of thousands of dollars in just dev time spent on trying to update it ( ignoring money lost because that dev time couldn't be used for new features and value adds ) 2. plan to upgrade to larger disks that have become cheaper by the time disk space could possibly be an issue caused by using up 32 bytes instead of 32 bits.
|
# ? Mar 6, 2017 20:34 |
|
also in the event that you actually did want to store a bitmask or something, binary is what you want to store it in, not a loving int.
|
# ? Mar 6, 2017 20:35 |
|
infernal machines posted:dude, have you heard of mongo? yeah, I heard it was bad
|
# ? Mar 6, 2017 20:36 |
|
mongo is web scale
|
# ? Mar 6, 2017 20:38 |
|
Shaggar posted:lol. like I'm gonna use oracle. what am i, made of money?? sql server licensing is lol and so is using windows server to run it for anything that requires uptime thank god microsoft loves linux and is finally porting it to a real server operating system at least
|
# ? Mar 6, 2017 20:47 |
|
uptime? what is this 1997? you know clustering is a thing in windows right?
|
# ? Mar 6, 2017 20:50 |
|
infernal machines posted:uptime? what is this 1997? loling at the matrix of windows and sql licenses required to do that
|
# ? Mar 6, 2017 20:52 |
|
mishaq posted:loling at the matrix of windows and sql licenses required to do that Shaggar posted:lol. like I'm gonna use oracle. what am i, made of money??
|
# ? Mar 6, 2017 20:55 |
|
mishaq posted:loling at the matrix of windows and sql licenses required to do that theres no reason not to have windows datacenter and sql enterprise so who cares
|
# ? Mar 6, 2017 21:00 |
|
Shaggar posted:also in the event that you actually did want to store a bitmask or something, binary is what you want to store it in, not a loving int. lol if you distinguish between types like that instead of seeing everything as a field of bits and dereferencing everywhere
|
# ? Mar 6, 2017 22:21 |
|
thanks, bits thits
|
# ? Mar 6, 2017 23:50 |
|
Sagebrush posted:lol if you distinguish between types like that instead of seeing everything as a field of bits and dereferencing everywhere I don't even see the code anymore
|
# ? Mar 6, 2017 23:53 |
|
people who store bitfields in databases are spergs that are disturbed by the idea of "wasting" bytes and have a fundamental misunderstanding of databases. database fields should be stored in a way the RDBMS can sort, filter, order, and update on them. bitfields are the opposite of this. and if the DB is going to build an index on the field, which it should if you're selecting on it, it's already going to use much more than a byte per row. bitfields also can't have rational triggers and can't be updated without read-modify-write. so if you see someone storing bitfields in a database, piss in their coffee
|
# ? Mar 7, 2017 05:43 |
|
someone mentioned booleans in the discussion area on the blog re: him explaining his retarded bitfield poo poo on TEAMBERRY DOT COM or whatever he basically was like 'LOL i bet youre storing the string 'false' in the DB get outta here ya fuckin noob'
|
# ? Mar 7, 2017 06:11 |
|
in college i had to take a "media arts programming" class (or something like that) where you had to do a variety of different projects on different topics, with your choice of programming language and platform one week the topic was "databases and data storage" and i didn't know poo poo all about real dbs so i made a two-axis "robot arm" out of some servomotors glued together, tapped the potentiometers, hooked it up to an arduino, and wrote some code to let you program a motion by swinging the arm around and recording the pot values, then play that motion back on the servos. like how those fancy industrial robots are programmed by having a human do the motions with the arm and then the robot mimics it. i had 512 bytes of EEPROM so i could store up to a 5-second motion path. i got an A on the project. to this day that's the extent of my "database" experience
|
# ? Mar 7, 2017 06:51 |
|
e: no that's not quite right, i did do a little bit of mysql to implement a word filter on a project where people were able to text in phrases that would be projected on a public wall. amazingly the filter stood up for almost 14 hours before people figured out how to sneak stuff past it
|
# ? Mar 7, 2017 06:52 |
|
no one cares
|
# ? Mar 7, 2017 07:34 |
|
The do nothing rickety shitbarge option is looking more appealing
|
# ? Mar 7, 2017 08:53 |
|
I still dont get what the difficulty is with porting the archives. like whats the big deal, that seems like maybe the most trivial part of the entire concept of launching on a new platform. i dont know poo poo about poo poo, help.
|
# ? Mar 7, 2017 09:39 |
|
ETL is the hardest problem in computer science and won't be solved for another decade
|
# ? Mar 7, 2017 13:45 |
|
the archives is like 95% garbage poo poo posts and what isn't is in the goldmines. the only reason anyone would reasonably see losing the archives as a show-stopper is if they wanted to be able to dig through people's post histories for more wank material for the offsites.
|
# ? Mar 7, 2017 14:27 |
|
Aren't the archives broken anyway and posts havent been getting archived for a couple years now?
|
# ? Mar 7, 2017 15:04 |
|
|
# ? May 15, 2024 04:41 |
|
Fergus Mac Roich posted:I still dont get what the difficulty is with porting the archives. like whats the big deal, that seems like maybe the most trivial part of the entire concept of launching on a new platform. i dont know poo poo about poo poo, help. iirc it's a combination of archived post syntax being scattered across lots of incremental changes, so a straight import would result in a ton of unreadable garbage mlyp
|
# ? Mar 7, 2017 18:52 |