| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9
Comments:
<0> iterator isn't that big
<0> anyway
<1> I see one const_cast
<1> and it's defined inside a BOOST_WORKAROUND(__BORLANDC__, ...
<2> gee, I wonder if it's for a "broken compiler"
<1> the other one looks like it's part of a traits cl***
<0> they are calling
<0> a non const function from a const function
<0> so it isn't a broken workaround compiler
<1> i don't see it
<0> but a working work around compiler
<0> you have just one deference function
<1> i searched the iterator library, and only found 2 occurances
<0> and it uses for const and non const...
<1> Pegazus you're going to have to find it and point it out
<0> you are not supposed to do that normally
<0> i.dereference()Access the value referred to
<0> you just implement dereference once
<0> and you use it for const and non const deference
<0> anyway, i can give you the same example, but clearer on a similar library if you want
<1> i have no idea what you're saying
<1> the bottom line is that casting is lying, and if you have a decent design, you rarely need to cast - and if you do need to, it's often indicitive of a design flaw
<0> you rarely need to cast --> rarely is very far away from never...
<1> who said never?
<0> vawjr
<2> I only see 2 in the iterator subdir, one is the workaround, the other in traits
<1> where?
<2> lvalue_trait or somesuch
<1> Pegazus quote what vawjr said
<0> "<2> as I said desigh/architecture problem" i don't see the world probably or likely there
<0> world = word :)
<1> <@vawjr> if you HAVE to cast, you've likely a design/architecture error somewhere
<0> you didn't said that in the next line
<2> I was commenting on the ****ed design of that function you were using
<1> don't pick apart words to make an argument
<0> if you have a function
<0> wich takes 20 arguments (for a good reason)
<1> 20? geez
<0> and you sometimes modify some and sometimes doesn't
<0> and library specification ***ures you that in certain conditions
<0> the function won't modify that thing
<0> and you are under those conditions
<0> is that a bad design?
<0> or should i write
<2> yes
<0> 2^20 overloads?
<2> the library is poorly designed
<0> so you don't need to cast them...
<0> good
<0> let's write 2^20 overloades
<0> because that's a good design
<0> with 1048576 times the same function
<1> look
<0> -_-
<1> if it takes 20 arguments and some or all may be modified, then none should be p***ed as const
<2> you're ***uming that the original (did you write it?) is a good design
<0> <1> and what if you need to p*** it const args?
<1> they shouldn't be const, if they may be modified
<1> see, it all goes back to a design flaw *somewhere*
<0> unders certain condigions (thay you are actually meeting) some of those aren't modified, and you know wich, and you only need to p*** those by const
<0> no
<1> it's quite possible that the design flaw is pervasive through your code - in which case, it's still a design flaw
<0> will yo do 1048576 overloads?
<0> tell me
<0> will you do them?
<1> no, i would fix the design
<0> how?
<0> tell me
<1> how the **** should I know?
<2> Pegazus 1) I challange your statement that a function w/ 20 arguments is a "good design"
<1> i don't know what the design is, ***hole
<0> "<1> how the **** should I know?" --> that's the point
<0> that's the same point vawjr told today about code guidelines
<1> how is that the point? that I can't see the code?
<0> there are always exceptions
<0> because you don't know
<0> all the c***es
<0> so
<2> 2) your ***umption that the only alternative is 2^20 overloads is equally foolish
<2> this is VERY LIIKLEY not one of them
<0> shut up saying it's always like that
<0> :)
<0> thanks
<2> we said likely
<0> you aren't
<2> you posit an pornographic function w/ NO description at all and say //// "see I need it"
<0> ?
<0> what?
<1> most likely the situation you're describing is a design flaw. we can't say yes or no because we have no idea what the function is, or does, or how it integrates with the rest of the system, or what in hell the system might be - but we can still say it's likely a bad design
<0> rdragon: that's the point,. you can't but you are saying it
<0> i accept you say it's likely
<0> i don't accept you say it's 100% sure
<0> that's my point
<1> and I didn't
<0> you did
<1> 99% sure
<0> in some sentences
<0> i accept the 99% sure and i agree with you
<0> but not 100%
<1> okay, then stop arguing about it
<0> you stop
<0> :)
<0> anyway
<2> the original quote said LIKLEY
<0> how the hell to make the read write mutex?
<1> and start fixing your crap code
<0> my code is good
<1> except all the parts that aren't
<2> you posit a rdiculous idea and I say "see bad" then you start jumping all over thee plae
<2> Pegazus you've yet to demonstrate that
<1> anyway, what problem are you having creating a RW mutex?
<0> i don't know how to do it properly :)
<1> i'd suggest building it off of boost::mutex
<0> my problem is
<0> how to grant writer can get reader right
<1> do you know how one works?
<2> I used to...but I haven't had to play w/ messed up OSs to do it
<0> only reader can get writer right
<2> we were _writing_ the OS when I did it
<0> and multi readers just one writer
<1> heh
<2> but not concurrently
<0> what?
<2> i.e. if 1 writer, NO readers
<0> i want that the writer can gain reader access too
<0> so it doesn't lock itself
<0> :)
<1> yes
<1> if you have a write lock, you should also be able to get read locks
<0> and if you are the only reader
<0> get writer too
<1> if you have a read lock, you should be able to get a write lock if no other threads are reading/writing
<2> you want the writer to be able to invoke the read mechanism of the lock?
<0> what?
<0> anyway, i was told there was an "Optimal" implementation with all of this
<0> but i've not been able to found it
<2> one hopes you have a reliable thread ID availalbe
<0> i don't get what you mean vawjr
<1> what does 'optimal' mean in this case?
<1> it's a pretty simple idea
<2> non-trivial to implement
<0> that does all of that
<0> without a container
<1> it shouldn't be -that- tough, with a decent way to get a thread id
<1> hm, without a container
<1> that might be tougher
<2> you likely need tighter communication w/ the OS than you have
<1> Pegazus do you HAVE something working that you want to make 'better', or are you at the beginning?
<0> beginning
<0> (ive mutex cl*** and all that anyway)
<1> then i'd say get it to work first, then try improving it
Return to
#c++
or
Go to some related
logs:
#javascript
javascript onclick kill session
www. 4asi. ro
sony hd camcorder h3c
wc3 sql
#teens
bootfare
wtfux bdsm
Xdccklipper.mrc
baxxilu
|
|