| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
<Return0> I made a static member
<Return0> that determines the default format to be returned
<Return0> esp. since you wouldn't change it often in all of your use of the matrix multiplies
<MachinShi> never cared for matrix cl***es ****ing overloading ops. find it evil, myself
<Return0> 8
<Yuhjn> see, machin agrees
<MachinShi> qed we're right, you're wrong. stfu and do it our way ;)
<Yuhjn> matrix_result = m1.multiply(m2);
<Yuhjn> that's generally what you see
<Yuhjn> or m1.dot(m2); m1.cross(m2);
<Yuhjn> etc
<MachinShi> eh.i'd rather do -> multiply ( m1, m2 ) ; but still.. ***ertion holds ;) or stick it a Math cl*** or wahtever
<MachinShi> what'd you do for cross anywa? % ? ;) i've heard ppl argue for that
<MachinShi> Math::multiply ( m1, m2 ) ;
<Yuhjn> or more like Matrix::multiply(m1,m2);
<Yuhjn> but right * is pretty ambigious
<Yuhjn> so is multiply() but that's not the point :)
<MachinShi> it's more descriptive
<Yuhjn> yes
<Return0> well the problem is actually the abiguity in the result
<MachinShi> and multiply_matrix is kinda lame. since you got m1 & m2 which are the types you're multiplying :)
<Return0> it can be either row major or column major?
<Return0> should I make it some optional parameter or something?
<Return0> or 2 different multiply functions
<MachinShi> Matrix<float, row> m1 ; //maybe?
<MachinShi> or make it an enum type you p*** in when you make the matrix objecdt
<MachinShi> rather Matrix<float, Matrix::row>
<Return0> errr
<Return0> but you do a multiply function you can't overload them by the return type
<Return0> so it would have to be a parameter for the multiply function
<MachinShi> well.. lemme ask you this. what is it mean for a matrix to be rowmajor or column major? can that be changed at runtime? can you multiply a rm-matirx to a cm-matrix?
<Return0> like m1.multiply(m2, Matrix::ROW_MAJOR);
<Return0> and it could be a default parameter I guess
<Return0> yes yes and yes
<Return0> it means the way it indexes basically
<MachinShi> is rowmajor vs. columnmajor an element of the matrix of an element of the mulitply operation?
<Return0> rm and cm are just transposed
<Return0> matrix.
<Return0> but it matters for the multiply
<Madsy_> I actually agree with MachinShi, Return0. You only need a simple transpose function.
<Madsy_> It will probably be faster too, and less confusing.
<Return0> yeah but it would be more efficient to index them directly
<Return0> based on the format, when I do the multiply
<MachinShi> why not have multiply delegate to a function that does the actual multiply based onthe formats? (a private function )
<Return0> yeah that's what I'm thinking
<Yuhjn> i recommend strategy patter (see gang of 4)
<Yuhjn> pattern
<Return0> or even if it just had a nested if, there's only a few possibility
<Return0> err possibilities
<Madsy_> Return0: Ok, I have another idea. Make a matrix base cl*** you derive from. Then you make two cl***es, column-major and row-major.
<Return0> I'm listening...
<MachinShi> no, ewwwh. that's silly
<Return0> I think a format flag works fine really...
<Madsy_> Derived from the base cl***. Then overload operator * for both cl***es to take itself as an argument, in addition the the other cl***.
<Return0> I can even do some index re***ignments symbolically
<MachinShi> it's not a type of matrix. it's an element of the matrix. you can change a column major to a row major matrix. how do you do that w/ inheritance?
<Return0> 48 ***ignment operations
<Return0> handles all cases
<Return0> all 8 cases
<MachinShi> *twinge*
<Madsy_> MachinShi: Hm, what about unions? :)
<Return0> heh :P
<MachinShi> unions? eh?
<Return0> if not that, I need to have the multiply code listed 8 times with different indexing each time
<Madsy_> Nevermind.
<Grundee> Quick question: How do I edit resource files in dev c++ 5? The help files are misleading and I think that they were from an older version.
<Return0> I think that's the most efficient
<Return0> it's all internal
<Return0> If I'm going to use a function for the multiply, it seems like it might even be best if I p*** in the address of the storage cl***
<Return0> or something.
<Yuhjn> i think you're putting too much thought into it.... just "make it work" .. then later when you need to extend the functionality or fix a bug, consider the design again and refactor if necessary
<Return0> I just think transposing a matrix before the multiply internally is inefficient when I can just access the array differently.
<Yuhjn> doing all the design up front never works
<Return0> well efficiency seems pretty important
<Yuhjn> effeciency is fixed last using your profiler
<Return0> for something with so many multiplies
<Grundee> Anyone know about devc++ resource scripts?
<Yuhjn> design should not be constrained by effeciency
<Return0> I agree
<Return0> well not the interface anyway
<Return0> but internally I think it should be.
<Return0> I'm a very meticulous coder =\, I think things through... I think it's a good attribute.
<MachinShi> you're not lazy enuf. tha's your problem
<Return0> heh :P
<MachinShi> perfectionism in a coder is a bad thing
<Return0> why?
<MachinShi> least if you wanna do *WORK* as a coder that is
<MachinShi> if you're gonig into academia who ****ign cares, right? :)
<Return0> well, I take pride in my code, I don't just write crap for some company I don't care about
<MachinShi> you miss my point
<Karius> Say that to anyone in any business
<MachinShi> i take pride in what i write too. what i realize however is that i will never have the time to work on the code sufficient to make me 100% happy.
<Return0> Karius, what do you mean?
<Symmetry_> MachinShi, yes
<czthwrk> MachinShi: bah, what you mean is almost anything taken to extremes is bad
<Symmetry_> true beauty comes if you do not seek beauty
<Return0> I don't really expect 100%, but I put a lot more time into things, I'm ok with it.
<czthwrk> my recent Net::SSH2 module makes me happy, but there will be more versions
<Symmetry_> as long as it uses design patterns, czthwrk =)
<Karius> Return0: The people who feed you want quick results, a short test phase and dirty fixes, then deployment. That's my coding life at least.
<czthwrk> Symmetry_: yeah. Foote and Yoder :P.
<Return0> yeah that might be a problem.
<Return0> obviously I'll have to make that compromise when the time demands it
<Return0> anyway ...
<Karius> I'm right now bluffing to my manager about doing 'fixes' will I actually redesign a whole system. That's another way of doing it :)
<Return0> while*?
<MachinShi> czthwrk: that's sufficiently good enuf to state :)
<Return0> I find when I put in a lot of thought into something and do a good job writing it, I rarely have to go back and fix it
<Return0> In reality I find I'm more productive coding than my peers, at least at my college.
<Return0> anyway, howd I get on this topic ><
<Return0> here I am just chatting... heh
<Genia4> guys, is beta 2 the latest vs 2k5 version?
<job> it's gone release
<job> but yes beta2 is the latest beta
<Genia4> it has?
<Genia4> so how come beta2 is still available?
<MachinShi> cause it's not "out" yet. you can only get if you're an msdn subscriber
<Genia4> job: ?
<Genia4> oooh
<Genia4> when will it be out?
<Genia4> out for real
<MachinShi> dunno. it's a full release, so it may not be available "for free" for a while
<kawfee> VS2005 has been released
<Genia4> how many cd's is it anyway?
<kawfee> dunno, it's on MSDN
<Genia4> previous one was like 7... (including msdn)
<kawfee> cd's or dvds?
<kawfee> Isn't everything being relased on dvd now?
<Genia4> I wonder what kind of dicount ms offers for student licenses...
<Genia4> I want a windows cd ;)
<Genia4> kawfee, 2k3 was released on both cd's and dvd
<kawfee> MS gives students a version of VS for no charge
<kawfee> it's an academic version
<Genia4> I need windows
<Genia4> too
<kawfee> they give them windows too
<kawfee> and office
<kawfee> and sql server
<kawfee> and this
<kawfee> and that too
<Genia4> for free?
<kawfee> they give them pretty much everything
<kawfee> yes
<Genia4> find it hard to believe
<kawfee> I get it for free
<Genia4> free for home use/
<Genia4> as in a cd?
<Genia4> with a nice MS hologram?
<kawfee> Genia4: maybe you didn't go to the right university then =]
<kawfee> no, not on CD
<kawfee> they allow you to download it
Return to c++
or
Go to some related
logs:
football
rock
politics
football
chatzone
error: "expression must have a constant value" com
|
|