image_error

[natirips]

btw they found the antivirus as the guilty in the end, at least i understood her like that.
so steve was funny half-right hihi

So anti-virus did more damage than a hypothetical virus would have? >_<


@Steve: I don't pay my studies , the state does (for the most part). However, I must admit I learnt almost nothing (about informatics) at the university (although I learnt a lot about physics). Most of what I know I know from personal experience.

[SvaRoX]

Quote:
| 1. right(you) --> quiet(you)
| 2. quiet(you) --> peaceful(me)
|----
|
| | 3. right(you) ^ not.right(me)
| |----
| | 4. right(you) ## ^ELIM 3
| | 5. quiet(you) ## -->ELIM 1,4
| | 6. peaceful(me) ## -->ELIM 2,5
|
| 7. (right(you) ^ not.right(me)) --> peaceful(me) ## -->INTRO 3-6

AI code ?

[SteveMcqueen]

no, a quick selfmade logic proof done with the fitch calculus, stuff i learn while i am on my way to a bachelor on information systems.
sure you can use this for ai code somehow in altered form.

well, this stuff and uml have things in common:
- both have its uses
- students are forced to learn it
- in worklife people just rarely actually use it...
while theoretically searching for better algorythms by hand and ultimately proofing their functionability. you get the idea, its the really freaky stuff.

[SvaRoX]

well, this stuff and uml have things in common:
- both have its uses
- students are forced to learn it
- in worklife people just rarely actually use it...
while theoretically searching for better algorythms by hand and ultimately proofing their functionability. you get the idea, its the really freaky stuff.

Ahah it reminds me so many hours spent in school to learn some ugly pseudocode and models made by teachers for teachers =) Funny, I used to hate uml in school but now I think I would use it as much as possible, at least for collaborative or "big" projects...

[natirips]

Well, so far I haven't used pseudo-code, diagrams, UMLs, etc. etc. even once unless I was forced by teachers to do so. I simply write the code. Perhaps pseudocode was very useful in 60s and 70s, but with modern text editors it's pointless, IMO at least. I find UMLs usefull only if your code is absolutely unreadable or if you're making a closed-source API; either case is disgusting to me.

Edit: oh and btw, I do understand most of Steve's pseudocode-thing. However I don't understand "ELIM", "-->" and "##".

[SvaRoX]

Perhaps pseudocode was very useful in 60s and 70s, but with modern text editors it's pointless, IMO at least.

Using pseudocode is mainly useful if you want to write an algorithm and focus on the algo itself rather than deal with language-specific syntax or workarounds. That's why in some schools it's the first thing you learn : no need to understand how OS or code execution or memory works, you just concentrate on the operations.

I find UMLs usefull only if your code is absolutely unreadable or if you're making a closed-source API;

Or open-source API ? People willing to participate to the project don't need to read all the code, just a glance at the beautiful model with beautiful rectangles and arrows is enough to understand it

[natirips]

Just read the headers/interfaces?

[SteveMcqueen]

it's better to use a pre-designed model that works, that way you avoid great deals of useless dead ends.

this is useless for small projects where you do not have a put a great amount of work designing the software architecture itself. there a good IDE is really all you need.

but you are completely fucked if you are a project coordinator with 10+ people all doing their own thing in their own IDE's. because they dont need anything else and are so f*k good at it. how to organize the chaos and make distributed programming really work?

problems arise as soon as you do not work in groups with less than 3 people anymore. There it already starts making sense to use diagramms of whatever type. Unified Modelling Language just happens to be standard.
in informatics it is the trick to make good systems which are easily understandable for everybody else in the team.
visualized through models which are to be implemented afterwards by the programming staff with good code.
just the programmers get the interface definitions in the end.


lets take another approach. the design team always exchanges a bunch of plaintext-interface definitions... for what? they dont need to program anything at all.
for example: goal specification happens in europe, implementation happens afterwards in asia because it is cheaper. only the asians have the need for the plaintext itself.

system design, software construction done engineer-wise in particular, is something which is done simply easier through UML (or comparatives), no matter if you in particular need it or not. others do.
it is simply quicker to understand compared to having all interfaces as plaintext.

digression:
lets abstract another layer...
system design per se needs models: what if you want to model i.e. a business system, like a sales company?
something you need no programming at all for in the first place? try this with some interfaces then....
even though meta-models suck, they are still the things all specifications derive from in the respectively used model.
models which can simply be used in program design, too...


@natirips:
svarox pointed out to this two posts before... sure you got the hint? as always you claim you perfectly know all this. even though you studied physics and not informatics.

you also really know what this is?
first order logic is not pseudo code. it happends to be found not just in informatics.
still it is useful for proofing two algorithms to be equal on a mathematical base. Similar things you show through inductive proofs for recursive functions.
Pseudo code is your try to implement what you found out, before actually writing it the specified programming language.

p.s.
talk about off-topic

[SvaRoX]

Yes headers and interfaces are parts of an implementation of the model. The solution to a problem is described by the model, which can programmed in many different languages, so the reference is the language-independent model, never the code.
Ok you don't need any model to write some nerdy low-level algorithm

Never heard of Fitch... do you have any example where it's efficient to use that ?

[SteveMcqueen]

fitch is a calculus. a calculus is a set of rules consisting out of some axioms, with with other conclusions can be reached.
you can prove euklid's geometry rules simply by logic, for example.

fitch calculus is something we have to learn in university in "math for informatics", it is part of our introduction to discrete math. fith calculus is simply part of our introduction to logic in particular.

in general we have "MfI" so we get the idea on how to realize underlying principles behind informally given structures. especially of ones with non-numerical character. so we can afterwards put them into math terms. which can afterwards be put into code. maybe this will help you a bit to get better idea, as i am taught pretty similar things like these described in the link.