IRC Logs

Network: freenode Channel: #csharp

	Enter your search terms Submit search form
	Web www.irclog.org

[23:20]<mjzymygo>wow... what a great way to throw huge amounts of security issues at a solution...
[23:21]<sdnffdxdlld>ello matt
[23:21]<mjzymygo>hi dude. sorry I couldn't chat earlier. My 'puter had to be used as the demo machine
[23:22]<sdnffdxdlld>tis cool :)
[23:23]<mjzymygo>uh... and the way it handles connections... has the author ever heard of M$ Patterns and practices?
[23:23]<mjzymygo>Smippy: what was the question again? You fired it at me but I didn't get to read it
[23:23]<sdnffdxdlld>its 2 parts
[23:23]<faccau_>Moridin8, was that in response to my SQL_CLI?
[23:24]<mjzymygo>um... shit
[23:24]<faccau_>heh
[23:24]<mjzymygo>yeh. sorry dude didn't realise it was yours.
[23:24]<mjzymygo>but what I said is true
[23:24]<sdnffdxdlld>part 1 ... i'm thinking of using an optimistic concurrency style check for populating previously gotten data to save on bandwidth. would that be more efficient (bearing in mind each row would have to be individually checked or would that take longer than sjust sending everything anyway
[23:24]<faccau_>no, tis fine, if you really beleive it is shitty, I want to know why
[23:25]<faccau_>were both the security and connections statements regarding it?
[23:25]<rnzrnf_->how could i create such easy accessible enums like e.g. socketflags?
[23:25]<sdnffdxdlld>Azrael_-, look up enumerations :P
[23:25]<mjzymygo>Lokkju_ I'll be happy to discuss, but smellyhippy asked me a question about 6 hours ago that I need to answer...
[23:26]<mjzymygo>smippy: how are you planning on hitting the DB? using datestamp ?
[23:26]<faccau_>Moridin8, np - just make sure you realize this is for a one time or batch type stuff, for using from the command line - it is not meant for anything else
[23:26]<sdnffdxdlld>timestamp
[23:26]<mjzymygo>It pretty much depends on what you need the data to do and what your sending.
[23:27]<mjzymygo>is the model similar to MR?
[23:27]<faccau_>Moridin8, and, it is meant to be able to use *any* SQL capable odbc/ole driver - I had to do thousands of inserts into a VFP DB from a flat text file, once per week, so...
[23:27]<awjzdd>sup homies
[23:27]<mjzymygo>(well... what was planned ;o)...)
[23:28]<mjzymygo>lukkju... I use Sql Management studio for all my batch needs ;)
[23:28]<sdnffdxdlld>say i have a collection of data objects and i need to refresh them to make sure they are still curent ... I could send over each of thier timestamps and check if any updates have been made ... if they have i can get the updated records ... but would the bandwidth (and time since its over adsl) be worth the cost of each individual hit per item/sub items
[23:28]<sdnffdxdlld>hey wormy
[23:28]<awjzdd>sup dewd
[23:28]<rnzrnf_->k, thx (didn´t know you could define the enums also outside of classes
[23:28]<mjzymygo>smippy: are you working on a LAN?
[23:29]<awjzdd>i am gonna rock out some monorail tonight
[23:29]<faccau_>Moridin8, SQL Management studio does not handle files that are over a million lines of commands very well - and, that requires you install it, which is a bit heavy for a lot of purposes
[23:29]<sdnffdxdlld>wan
[23:29]<sdnffdxdlld>if it was lan i'd just whore the bandwidth :P
[23:31]<mjzymyrnyrxv>sodding internet
[23:31]<sdnffdxdlld>:/
[23:31]<mjzymyrnyrxv>Where were we?
[23:31]<sdnffdxdlld>[21:28:55] <smellyhippy> wan
[23:31]<sdnffdxdlld>[21:27:12] <jwormy> sup homies
[23:31]<sdnffdxdlld>opps
[23:31]<sdnffdxdlld>[21:29:08] <smellyhippy> if it was lan i'd just whore the bandwidth :P
[23:31]<awjzdd>LOL
[23:32]<sdnffdxdlld>how the hell did i copy your texte!
[23:32]<sdnffdxdlld>it;'s miles above!
[23:32]<mjzymyrnyrxv>If your dealing with the internet I will always ask this (and you will probably know these questions)...
[23:32]<mjzymyrnyrxv>1. How large are the datasets (rows)
[23:32]<mjzymyrnyrxv>2. How large are the column sets
[23:33]<mjzymyrnyrxv>3. how often are checking back to the data source
[23:33]<sdnffdxdlld>reads are done with a datareader and potentially up to 100k of lines (but I'm sure we'll introduce paging)
[23:33]<mjzymyrnyrxv>4. how secure does it need to be
[23:33]<mjzymyrnyrxv>I'd heavily suggest paging going in to start with
[23:33]<mjzymyrnyrxv>remember the paging controls for ACMS?
[23:33]<sdnffdxdlld>just about
[23:34]<sdnffdxdlld>I know how to impliment it but i need to discuss the details at work
[23:34]<mjzymyrnyrxv>ah
[23:34]<sdnffdxdlld>(the director says we dont need a spec lol)
[23:35]<mjzymyrnyrxv>100k... thats huge in a way. whats the allowable response time?
[23:35]<mjzymyrnyrxv>ah.. no spec... ignore above question
[23:35]<sdnffdxdlld>lol
[23:35]<mjzymyrnyrxv>in that case i'd just implement it the best way you know how in as short a time as possible and blame the lack of a spec when it fucks
[23:35]<mjzymyrnyrxv>;)
[23:36]<sdnffdxdlld>aye ;)
[23:36]<mjzymyrnyrxv>I'd go with paging and opt-con date checking.
[23:36]<mjzymyrnyrxv>sod the band-width
[23:36]<mjzymyrnyrxv>if the boss doesn't have a feasability study on this shit then it's his issue.
[23:37]<mjzymyrnyrxv>if you remember I had to force our mutual ex-employer to do that kinda shit. Imagine how the councils/ACMS debacle would have panned out if I hadn't.
[23:37]<mjzymyrnyrxv>I'm not sure if i've clarified your query though... *ahem*
[23:38]<sdnffdxdlld>I might see if I can do somethign funky with triggers, a comm server and notifications of changes made
[23:38]<mjzymyrnyrxv>what would you want the triggers for?
[23:38]<sdnffdxdlld>that would remove the need for checking for changes and save even more bandwidth hehe
[23:38]<sdnffdxdlld>true i could just do it straight from the stored proc
[23:39]<mjzymyrnyrxv>exactly.
[23:39]<sdnffdxdlld>but with triggers it would fire if any manual changes were made wouldnt it?
[23:39]<mjzymyrnyrxv>if bandwidth is a total premium, look into doing data transfer using XMLHttp or use remoting.
[23:40]<mjzymyrnyrxv>welll yeh. unless the person has the permissions to turn trigger firing off
[23:40]<sdnffdxdlld>I have a few prototypes and palns for that knocking around just in case ;)
[23:40]<mjzymyrnyrxv>well. be careful with triggers. they can play silly-buggers with transactions when playing across scopes
[23:41]<sdnffdxdlld>yeah but the person manually editing would be one of the dev team ... it means live changes would refresh the connected clients
[23:41]<mjzymyrnyrxv>I'd stick with the stored proc's if no-one at your comapny understands SQL server with any savvy...
[23:42]<mjzymyrnyrxv>well... if your using the triggers to enforce rules on firehose, dont forget check constraints.
[23:42]<sdnffdxdlld>how do they mess with the transatcions?
[23:42]<sdnffdxdlld>and yer the constraints are there from the start
[23:43]<mjzymyrnyrxv>You know how I insist on placing even single write statements (update/delete/insert) within a micro SQL transaction even though 'technically' they aren't needed?
[23:43]<sdnffdxdlld>oh apart they had to be down gradedto allow 0 as an acceptable value :/
[23:43]<sdnffdxdlld>yer
[23:44]<mjzymyrnyrxv>Well thats in case some bozo decided to ruin the design model and through triggers against the system (a-la my predacessor at our mutual ex-employer).
[23:44]<mjzymyrnyrxv>You can get triggers to dump transactions and/or do extra work other than initial single write statements.
[23:44]<mjzymyrnyrxv>it gets messy, but its about protection.
[23:45]<mjzymyrnyrxv>I have a proof case here somewhere I did for my current client to prove why something was dying.
[23:45]<sdnffdxdlld>yer, these triggers would just be used to fire a message to the comms server though saying "x has changed with x ID
[23:45]<sdnffdxdlld>"
[23:45]<mjzymyrnyrxv>I'll try and find it out if I still have it for you and email it you tomorrow or something
[23:45]<sdnffdxdlld>which would then be sent to the connected clients
[23:46]<kjr>hey now...
[23:46]<sdnffdxdlld>;)
[23:46]<mjzymyrnyrxv>hiya kog...
[23:46]<kjr>moridineight: hi
[23:46]<mjzymyrnyrxv>cool.

Back to #csharp
Or go to some related logs:
#debian
#gentoo
#debian
#blender
#iptables