@RobertHarvey Why? Usually subqueries are nothing to shy away from. They may get a little odd sometimes, but they're usually not particularly large - any given SQL query only has a small number of clauses to it, subqueries no different, I've never seen someone trying to avoid them for reasons other than performance. I've never had difficulty maintaining them nor seen anyone else either... CTEs though do give syntactic sugar to make them clean. You can use a view if really necessary

Some of my earlier Access programs had three layers of subqueries. It was a batch process having 30 or so queries executed in four or five steps (for an MRP system); it worked surprisingly well.

Eventually I had to break it up into intermediate results tables.

But each query had a name. SELECT queries in Access are essentially Views.

The fact that Java's Observable is a class and not an interface upsets me.

composition over inheritance?

jk that is older than that

I wanted to make a parent type that had a collection of empty methods for different types of messages, then each child would choose the message(s) it cared about and provide an implementation for those. Although...I did miss the obvious solution. The parent class extends Observable and the children classes just extend the parent.

Dammit, @Oded. You're too damn fast.

39 seconds from post to close...

he has to be refreshing the page constantly

My record is in the 30s, too.

This is a non-trivial comparison operation, I think. Unless I'm overthinking it. My timestamp is broken up into an integer microseconds component and an integer seconds component.

I've got to be overthinking this one...

@GlenH7 It's not my idea. It was the raw packet data that was implemented that way. My POJO to hold raw packets follows the same structure.

if(this.sec!=other.sec) //compare micros

Yeah. That's what I have now. Compare the seconds. If they are the same, look at the microseconds.

or combine into a single long first

@ratchetfreak naw. Just happen to see the tab getting a number and seeing what it is about :)

RealTime FTW

micro is only 6 orders of magnitude

Wait...wtf at this spec. The microseconds "may be greater than or equal to 1,000,000". Isn't 1,000,000 microseconds = 1 second?

maybe there is no 0

@GlenH7 that was my siggestion

I think just adding to a table that the user polls

:15134683 web sockets with fallback to XHR long polling

:)

one is just pinging servers all day the other is a pain

What do you think about the term proximity programming.

One step from pair programming, sit within 5 m of the devs one works with.

I think it could solve many problems.

isn't that just status quo for many people?

@GlenH7 communication, flow of info (same thing)

distance is def a factor

don't think relying on meetings is the way to go

yeah, we don't have a good room as it is now

IM trumps closeness most of the time

I bought a GoToMeeting accout but the chat client was so poor we deemed it useless

there is lynk, skype an friends

We have Notes at work and the Sametime chat client is nice

problem is that the consultants do not have access to it

if all else fails IRC

proxy blocks skype and many of the rest

Google Hangouts?

yeah lync can accept external phone calls

also we will not be allowed to discuss much on something that is not hosted by the company

you can self host the server for lync I beleive

yes good suggestions, ty sirs.

does it make sense tyo schedule cleaning for one whole sprint?

I'd say blurting on IM is not much of a problem, IRL or phone is much more intrusive that way ime.

and much faster

and less dead tree usage

yeah IM is great

I still think proximity has merit

just in same building is enough if IRL meeting is needed

I'd say ideally sit next to each other

if you have a IM with screen sharing, that will also help

perhaps even pair program.

Maybe not pair on everything, leave it up to the devs.

Pairprogramming depends much on the two involved ime

pair programming requires both devs to work on the same thing while unerstanding each other

yeah forcing is probably not good at all

compiler question

/parser

there's a 5 minute time limit on it I think

but you can always delete

I think it is two minutes for edit

when parsing a string, and replacing all, say, "=", "+" and "," characters with " = ", " + " and " , ", respectively. I want to keep track of the original position in the string (while iterating char by char, and the 'replaced' string, so I can give feedback to the user about where errors are (char pos. and line)

don't think you can delete forever

well i tried to edit it and can't so that t must be the limit there

what language?

any really - vb.net

super instead of storing the token store a token-position pair

c# will do

regex gives position for matches in C# I think

but i am using string.replace("string-to-be-replaced", "replacement")

it's often easier to just replace the string with a token stream that you don't have to do string operations on

so i can't keep track of positions.

my question could be read as, is it possible to use this function, and STILL calculate the original position?

not unless you keep the original or find the positions first

but I'll repeat preparse the string into a token array

i am keeping string orig_code and string altered_code

so i am keeping acopy

i was going to replace all " "(two spaces), with " " for now. and replace all "=" with " = " etc. to make parsing easier

my parser is my lexer xD

it's much easier to reason about a token stream than a string of chars

i think

lol

lexer just translates the string into tokens

the parser does the hard parts

i am trying to translate to tokens

but first i need to have chars like "=" sep. from the identifiers etc

i find it hard otherwise

that is easy in a lexer

i know

that's why i need them separate

they are diffzerent tokens but tokens non the less

how would i go about recognising "int abc=23443;" as "type:int id:abc op:= expr:23443 ?

a string "foo=bar+fuz" would then translate into tokens IDENT(foo), EQUAL, IDENT(bar), PLUS, IDENT(fuz)

without confusing abc=23443; as one string or word

it's easier to let all identifiers (type and variable names) as the IDENT token

when the lexer sees the = it knows that it is not part of the previous token

so is it possible to iterate source code char by char?

yes

or would it be better to split into tokens right off the bat

charAt in most languages

well java based languages

but similar enough

@GlenH7 aren't you always out of them around this time of day?

@MichaelT wouldn't that be very slow on large portions of code?

@MichaelT I was about to say – characters != bytes. As it happens, Java's characters aren't true Unicode code points either, so it isn't actually much better.

because i'd have to iterate them too

if anyone wants to look at my syntax, to see if i'm on the right track there?

it's like a grammar list

pastebin?

first get a substring of the interesting part (usually line by line)

pastebin.com/geBgatft SYNTAX of my language "Lupe"

remember the goal of the lexer (what we are trying to get you to build) is to take the input and create a token stream, it should be a deterministic finite state machine

it's a toy language which might actually work out (hopefully)

it's only in alpha, if you like

but the satisfaction of writing your own little compiler tho

..

if it works out i could port it somehow

anyone looked at my syntax, could it work?

your syntax expects individual chars for tokens

?

it's in VERY early stages, and i probably don't know that much tbh

@rachet how does it?

each <fgsdfsdffs> only represents an already defined code element

look at how you defined the <identifier> , a lexer would emit the identifier token dierctly

like: <type> := <int> | <string> | <char> | <bool>

this text file describes the language to ME. to help me write the code

a compiler shall never see it

letter and digit and all the <op> look for the actual characters

it is a visual definition of the language

yet there are tools that will transform that into a parser

such as?

yacc

but even then,it would limit my interaction with backend, no?

wouldn't it be rather easy to write a simple tokenizer

?

writing your own parser is rather complex

and solved in the general case

but if it only spits out AST then i can' control how certain things will be evaluated in special cases

can antlr do this (turn code into ast) from my syntx definiton>

whoa whoa whoa

alright everybody just calm down

all you need is a monad

Thankyouverymuch.

I knew you were going to say that

that or haskell

@MichaelT o yz, I can haz links for him!

don't scare him off

@ratchetfreak I'm not scary ...

your link flood is intimidating

@ratchetfreak my link flood is awesome, it's not my fault awesomeness intimidates you

all i want is a way to convert "int my_int=543;" into "[TYPE:=int] [ID:=my_int] [OP:==] [EXPR:=543] [SEP:=;]

??

@SuperCookie47 You just trying to get ASTs or something?

@MichaelT ...come now, in all honesty, my approaches would say otherwise depending on what he wants, my approach may actually be valid for this dude

if i get the above, i can then semantically analyse it, create ast etc.

@SuperCookie47 you can transform it into [ID:=int] [ID:=my_int] [OP:==] [INT:=543] [SEP:=;]

int is not an indentifier

and let the parser take care of differentiating which ID is a type or variable

and an int, will be recognised under expr

it's what nearly all C-style languages do

during lexing things are operators, keywords or identifiers

and the parser makes the decision about which should be which

but i can go from here straight to sem. analysis, then gen. ast, then code gen

for my simple purposes anyway

the ASt will be able to say that it is a type though

plus my types will never conflict with indent.s

no user defined types?

that will be later, IF i get it working

:)

if you want to ensure that they don't clash you need to enforce caseness or add a $ or something to the vars

they make parsing much easier don't they

but shant clash because it's case sensitive and they aren't allowed to be named types

for example int int; will never be allowed

plus what i am asking is how can i differentiate between my_int=234;

but that would get you into trouble when a new type is introduced and it clashes with a existing variable

how could i change that to $my_int = 234 ;

at least

but still keep track of origional parsing position within source code (unmodified)