Design questions c++, stl, mysql, fedora, 64 bit etc etc

[Mike]

Hello,

I would like some comments on my design ideas regarding this process I
intend to create.

I have a c++ program , let calls it NumberCruncher that I wrote on windows.
Its not a windows program but vanilla c++ utilizing stlport containers
(maps, hash table etc). I'm running into issues where the numbers I am
calculating are bigger than the largest data size currently available on
windows thus the idea to port to 64 bit linux.

There are other issues where I run out of memory also but the main problem
is the data size. My plans are to build a 64 bit cpu machine, load 64 bit
linux on it (I'm thinking fedora) and use gcc or g++ (not sure which) to
port my program over.

That sounds pretty clear cut but I realize there will be issues here and
there. I dont know but I'm assuming there is a 64 bit stl available.

Next I'd like to set this up so that my web machine (which is windows) ftp's
the data file over to the linux machine for processing. He then processes
the data file, loads the data into mysql then skims off the topX data pieces
I'm interested in and ftp's that back over to the windows side of things for
storage in sql server and display on the web.

Again sounds pretty clear cut but we'll see.

On Linux how would it detect that there is a new file available for
processing? Is that a cron job which watches the disk and how would it know
the ftp is actually complete and not in process, meaning the data is still
being transfered? The data files themselves are small so it might not be an
issue but given a chance for a problem a process will find it.

Thats about it. I'd just like some thoughts or ideas regarding this plan. My
background is c++ on windows so this will be my first real serious adventure
in to development on linux. When I think c++ on linux I think all command
line so what are the popular graphical environments to program in ?

I'm setting up fedora now on one of my dev machines, dual boot win2000 to
get a start on development, this isn't 64 bit yet.

Thanks alot!!

[Scott Lacy Salley]

Post by Mike
Hello,
I would like some comments on my design ideas regarding this process I
intend to create.
I have a c++ program , let calls it NumberCruncher that I wrote on windows.
Its not a windows program but vanilla c++ utilizing stlport containers
(maps, hash table etc). I'm running into issues where the numbers I am
calculating are bigger than the largest data size currently available on
windows thus the idea to port to 64 bit linux.

I may be wrong:

The general reason to use a 64 bit computer is to handle large quantities
of data, since the pointer size is 64 bits.

I think you'd be better off re-writing a bit of code to use a 64-bit type
(like __int64) on Windows before investing in a bunch of new
hardware/software. No matter what, you are going to have to re-write
application code to use 64-bit types.

--
Scott Lacy Salley
formerly Scott Lacy Smith

[none]

Post by Mike
Hello,
I would like some comments on my design ideas regarding this process I
intend to create.
I have a c++ program , let calls it NumberCruncher that I wrote on windows.
Its not a windows program but vanilla c++ utilizing stlport containers
(maps, hash table etc). I'm running into issues where the numbers I am
calculating are bigger than the largest data size currently available on
windows thus the idea to port to 64 bit linux.
There are other issues where I run out of memory also but the main problem
is the data size. My plans are to build a 64 bit cpu machine, load 64 bit
linux on it (I'm thinking fedora) and use gcc or g++ (not sure which) to
port my program over.
That sounds pretty clear cut but I realize there will be issues here and
there. I dont know but I'm assuming there is a 64 bit stl available.
Next I'd like to set this up so that my web machine (which is windows) ftp's
the data file over to the linux machine for processing. He then processes
the data file, loads the data into mysql then skims off the topX data pieces
I'm interested in and ftp's that back over to the windows side of things for
storage in sql server and display on the web.
Again sounds pretty clear cut but we'll see.
On Linux how would it detect that there is a new file available for
processing? Is that a cron job which watches the disk and how would it know
the ftp is actually complete and not in process, meaning the data is still
being transfered? The data files themselves are small so it might not be an
issue but given a chance for a problem a process will find it.
Thats about it. I'd just like some thoughts or ideas regarding this plan. My
background is c++ on windows so this will be my first real serious adventure
in to development on linux. When I think c++ on linux I think all command
line so what are the popular graphical environments to program in ?
I'm setting up fedora now on one of my dev machines, dual boot win2000 to
get a start on development, this isn't 64 bit yet.
Thanks alot!!

* Read this article about "Porting Linux applications to 64-bit systems"
http://www-128.ibm.com/developerworks/library/l-port64.html
---------

* Use similar typedefs and names (s32, u32, s64, etc) as used in the
Linux-kernel development. Take a look at this header file.
http://lxr.linux.no/source/include/asm-x86_64/types.h

38 ...
39 typedef signed short s16;
40 typedef unsigned short u16;
41
42 typedef signed int s32;
43 typedef unsigned int u32;
44
45 typedef signed long long s64;
46 typedef unsigned long long u64;
---------

Two sample files that employ STL.
http://www.futuredesktop.org/tmp/test-std1.cpp
http://www.futuredesktop.org/tmp/test-std2.cpp
---------

* Essential Fedora Core 5 guides
http://www.futuredesktop.com/fedora-core-5.html

That's all.
;)

// moma
http://www.futuredesktop.com/AsteriskPBX.html