| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12
Comments:
<0> Such an ugly read
<1> I don't know what he exactly wants to do
<2> send binary data in an http request
<1> how big is the data
<2> a jpg
<2> anywhere from 10k-300k
<0> vector<char>++
<1> use an std::vector
<1> er, what Ashe` said
<2> it has to be binary safe though
<0> Binary safe doesn't mean anything
<2> it cant terminate on \0's
<0> Why would it terminate on \0's
<1> ?
<1> neither an std::string nor an std::vector would "terminate on \0's"
<1> they will store what you tell them to store
<3> JBlitzen: with std::copy and istream iterators, you can
<4> Tell holospoof
<2> i will paypal someone $10 to write this code for me
<2> lol
<3> prices = Answers are a dollar. Answers which require thought are five dollars. Correct answers are twenty dollars. Dumb looks and newbie torment come free, of course. Homework done for $1500-$2000/day or any part of that day (What's an 'A' worth to you?). Think Paypal!
<2> im waaaay too tired
<1> holospoof wtf, we're telling you
<1> do you know about std::copy?
<1> and std::istreambuf_iterator?
<3> he doesn't need that, anyway
<4> Would someone just tell him what he does need already
<2> ill figure it out
<2> thanks for helpin
<3> JBlitzen: already did, on efnet
<4> Cool
<4> Thanks, kniht
<5> man, the STL are like copy maniacs or something. A single call to map's insert() generate 2 calls to copy ctor instead of 1 as one would expect
<4> Heh
<5> i p*** a pair to insert that I created and that has already cost me a copy call, and what it does with it? it creates another copy of that pair
<3> what do you expect? it has to copy it into the map, that's 1 copy, the other copy you did yourself
<5> no
<5> it does two copy calls on the insert alone
<5> setProperty(const CProperty& prop){pair<string, CProperty> thisKeyProp(prop.getKey(), prop);m_propLookupTable.insert(thisKeyProp);} generate three calls to the copy ctor of prop
<5> pair<string, CProperty> thisKeyProp(prop.getKey(), prop) <--- 1 call to prop's copy ctor
<5> m_propLookupTable.insert(thisKeyProp); <---- 2 calls
<5> for some reason, it recreate a pair from the pair I gave it
<3> then your implementation ****s
<5> my implementation is only two lines of code
<3> your std::map implementation is very likely more than two lines of code
<3> map::insert requires at least one copy (to insert into the map), but no more
<5> ok, you meant 'my' as in my compiler's?
<3> yes
<5> ok
<3> although they aren't inseparable
<5> yeah, it should only generate one copy on insert()
<5> but for some reason, the copy is done here first:
<5> template<cl*** _Other1,cl*** _Other2>
<5> pair(const pair<_Other1, _Other2>& _Right)
<5> : first(_Right.first), second(_Right.second)
<5> {// construct from compatible pair}
<3> however, this presumes pair<string, CProperty> is your value_type, and it's not
<5> second(_Right.second) is where the call to copy ctor happens
<3> so that's where the other copy is from, constructing a temporary to bind the map::value_type const& to
<3> map provides a typedef for a reason, use it
<5> sorry Kniht, I am not very efficient yet with STL. how can I use that typedef?
<3> I wouldn't bother for types that are cheaply default constructable
<3> m[key] = value
<3> Map::value_type is the typedef, where Map is the type of your map
<3> and I take that back, your implementation is fine, it's your use that ****s :P
<5> You bastard!
<5> I was happy there for a while that I did somthing that sisnt ****
<3> why did you avoid the obvious m[key] = value?
<5> because I benchmarked it
<5> turns out it is 4-5x slower then insert()
<3> your benchmark led you astray in this case
<3> and it would only be valid your implementation, in any case
<3> did you profile? is that your bottleneck?
<5> well, I will still use the insert() method knowing that it is 4x ~ 5x faster
<6> likoo: how to benchmark a c++ function?
<5> yes, on my implementation
<3> doesn't matter if it's 100 times faster, if that's not your bottleneck
<5> javaq:precisiom timers
<6> what software?
<0> He counted on his fingers
<5> Kniht:it could become a bottlneck though
<6> arrays is faster than vector
<0> Of course I wouldn't use it on the STL
<6> no s
<5> DevPartner is pretty good
<5> numega used to make high grade software, until they got baught by compuware
<5> I am still trying to figure out how to use that ::value_type
<0> map<K,T>::value_type
<5> yes, then what?
<0> insert( map<K,T>::value_type( key, value ) );
<5> oh, instead of pair?
<5> neat
<6> push_back instead of insert
<5> on a map? i though that wouldnt be good
<5> if the key exists it will override it
<5> but not at the back
<5> yay!! ::value_type saved me from that dreaded copy ctor call!
<3> likoo: anything could become a bottleneck, worry about known bottlenecks first
<3> and if you don't know any, find some
<5> ok, I think I worry too much about bottleneck all the time. Coming from C/Asm background, I don't even want to look at the code that the STL produces
<0> Even in C, you don't look for bottlenecks ever 5 seconds, just get it to work first
<0> +y
<5> I do :-/
<0> Yes, but that's cause you're bad ;)
<5> ever function I write has to be the most optimal it could be. and the I re-ietarate after the initial stage
<5> Ashe: lol, no, I am paranoid
<0> So it takes you 500% of the time to write it, but you gain 1% in speed
<0> Maybe it's the time you spend that you should make optimal, not the time your program spends ;)
<5> you see, for example, this ::value_type issue, I would have never knew about it. and GOD knows how many STL 'faux pas' i am making that are costing me speed
<3> and spending that much wasted time means he either misses more important improvements or he misses the schedule
<3> you don't have an STL reference?
<3> ref = Some C++ Reference Sources: (1) Your compiler documentation (2) http://www.dinkumware.com/manuals/reader.aspx?lib=cpp (3) http://msdn.microsoft.com/library/en-us/vcstdlib/html/vcoriStandardCLibraryReference.asp?frame=true (4) http://www.sgi.com/tech/stl/ (5) http://www.cppreference.com (*) http://www.google.com
<0> It takes you so long to write it that computers are 3x faster by the time you (if you ever do) complete the job
<5> Ashe:LOL
<3> you didn't know the parameter types for insert?
<5> well at work I code in JAVA, and I don't give anywhere near as much attention to speed then when I code at home
<5> Java is inheritly slow, and there isn't much I can do about it
<5> Kniht: I thought it took a pair, that is why I created one before hand
<3> what made you think it took a pair?
<3> guess = Programming by guessing does not work, so you better -->RTFM.
<5> I think some example I see on the net
<0> + by not using value_type your code breaks if you change the map
<0> (change its type, that is)
<7> a lot of examples actually use pair
<7> sadly
<3> 90% of anything is crud
<6> map<K, V>? k can be an int?
<0> Why not
<0> As long as you can do < on it
<6> k
<3> or use whichever comparator the map uses :P
<5> Kniht: crud? as in Create, Regroup, Update, Delete?
<3> no, as in ****
<1> haha
<5> lol
<5> HEY!
<5> I did read the RTFM, in fact this is where I got the samples from
<5> http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vcstdlib/html/vcoriStandardCLibraryReference.asp
<0> ****? as in Serialize, Highlight, Iterate, Treat?
<3> even though that's listed in the ref calc, I don't like it
<5> m1.insert ( Int_Pair ( 4, 40 ) );
<3> dinkum is better, and I use the standard itself
<5> ok
<3> Ashe`: no, as in SHut-up Internet Troll :P
<0> :'(
<3> Silence Hubristic Internet Troll just didn't work as well
<0> Sig Hup
<5> hmm..I am looking in map.insert on dinkum but I would have never guess what it needs or especially that I could p*** it a ::value_type
<5> pair<iterator, bool> insert(const value_type& val);
<5> Oh
Return to
#c++
or
Go to some related
logs:
#c++
#linux
#c++
lebcafe
glut.h in knoppix forum
is it expired
#chatzone
zenn diagram
#c++
lizbjani
|
|