|
I posted my experiences when porting an application here [^].
Maybe it is of some value to you  .
|
|
|
|
|
Every little bit helps
______________________
Mr Griffin, eleventy billion is not a number...
|
|
|
|
|
how can I make activex (.ocx) file
in c#?
I have a software that can get .ocx files
and not .dll files
|
|
|
|
|
You can't. C#, VB.NET and the other managed-code languages can't target building an old .OCX file. It can still be done using C++ though. The managed languages can "emulate" a COM component by COM-interop, but they cannot build an exact binary replacement for them.
Dave Kreskowiak
Microsoft MVP - Visual Basic
|
|
|
|
|
Hi gurus!
I’m creating a large array that holds the distances from all points in a location file (x and y coordinates) to all other points. I have no control over how many locations the user will input (thousands most likely) so with the number of distances to record at (n(n-1)/2), I’m going to be storing a helluva lot of values. As I want to use these in a Monte Carlo simulation and access each value up to 999 times, I would prefer to calculate the distance matrix only once and not have to recalculate each distance every iteration of the MC simulation.
I would appreciate advice on the most efficient way to store and quickly access these values: ArrayList, Hashtable, Queue or something else?
 A secondary question. Is what I am trying to achieve just unmanageable? For example, with the user including 10,000 locations, that creates 49 million distance calculations. How will this affect access time to reading the values again and again…?
TIA
Jerry
|
|
|
|
|
Hi,
if your distances are theoretical, why not just store city coordinates, and calculate
the distance every time you need it?
If it is actual distances, you will need to store of course; maybe a random-access
file on disk is appropriate.
In any case, I would be worried more about elpased time for your program,
when it will deal with 10000 cities visited 999 times...
I hope it is not a brute force attempt to the travelling salesman problem !
Luc Pattyn
|
|
|
|
|
I'm afraid I'm not familiar with the travelling salesman problem, so it isn't that!
The Monte Carlo simulation is one that keeps the spatial locations constant, but randomises a third variable (the first two being x and y location), and yes, these are real locations. My concern with recalculating the distances every time is that with 10,000 locations, that creates 49 million distance calculations. With 999 iterations of the Monte Carlo simulation (1000 with the observed outcome), that means that to complete the program requires 1,000 x 49 million distance calculations  . Hence my desire to only do the 49 million calculations once and then store the results in an array that I access the remaining 999 times.
Do you (or anyone else out there) think that recalculating the distances every iteration for 999 times will be quicker than storing the distance results in an array or a file and accessing them repeatedly?
Jerry
|
|
|
|
|
I still dont know what you want to do in each run, e.g. I am not sure you need
all the distances everytime.
If each run only needs "a few", do lazy calculation (i.e. only what is needed),
if you need all of them every time, I would go for the (large) file on disk,
which will automatically get cached when your code reads the relevant entries.
Also, how complex is one iteration ? is it much larger than calculating the
distances it needs, or not ?
Can you be more specific ?
Luc Pattyn
|
|
|
|
|
I need every distance, every time. The program is designed to conduct a new statistical process on locations that have a date attached. Think something like bird spotting. I have, say 10,000 records with an x and y coordinate and the date of observation. The program measures the distance between each point to every other point, and also measures the number of days between each observation. It then randomizes the day, and repeats the process 999 times, randomizing the days each time. The distance between points therefore remains constant and does not need to be recalculated, but it does need to be remembered every one of the 999 iterations, becuase it categorizes the results by distance. And there will be 49 million distances to remember.
The question therefore remains, for speed do I recalculate the distance every time on the fly, or save in a file, or save in an arrayList, Hashtable or Queue...?
|
|
|
|
|
OK, I start to get the picture.
Your basic information consists of some 10000 points (that could be 80KB of data).
So it seems very wasteful to expand that to a full distance matrix (some 50MB).
Let us assume that distance simply equals SQRT(dx*dx+dy*dy).
I did a simple experiment: the following code
double z=0.0;
for (int dx=0; dx<10000; dx++) {
for (int dy=0; dy<10000; dy++) {
z+=Math.Sqrt(dx*dx+dy*dy);
}
}
Console.WriteLine(z);
took less than 4 seconds on a 1.7GHz Pentium M.
Using a file on disk would certainly be faster (A large part of your file would
typically be in the file cache in memory, which gets refilled at the disk's
transfer rate, typically 100 MB/sec or more). So I expect consuming 50MB would take
under 1 second (you could try that with another 50 lines of code...)
On the other hand, the (lack of) speed of the above code is mainly due to the
Math.Sqrt() method; if we replace that line by z+=dx*dx+dy*dy; it runs in less
than 0.7 seconds.
Preliminary conclusions:
- if you are lucky and only need the square of the distances, go for direct calculation;
- if you are unfortunate and need a more complex calculation for distances (the earth
being round, remember), go for the file on disk.
- it may be worthwhile to investigate faster SQRT methods (maybe int based, maybe
slightly less accurate, whatever).
My suggestion:
I would certainly start of with direct calculation and see how it is behaving
before introducing a file-based optimization (the rule is: think first, then code
and get it working without unnecessary complications, then observe and
only then make changes, and introduce complexity, to improve).
Luc Pattyn
|
|
|
|
|
This helps a lot, thanks! 
I hadn't considered that the SQRT function was so time consuming, but given the iterations that I have, that is a big factor. I'll try what you suggest.
Cheers.
|
|
|
|
|
Hi,
I'd like to write a basic parser in XML to write the values of various fields across several forms to a text file. The idea is that custom settings of all these fields can be stored as a "user config" that can be recalled.
Could you help point me towards information online that will help me get started? I would like to start with a test app that writes several values to a text file and then reads then back into the form.
Thanks!
|
|
|
|
|
See this[^] set of tutorials.
/ravi
|
|
|
|
|
Thank you! Do you know of a similar set of tutorials for C#?
|
|
|
|
|
The tutorial supports both VB and C#.
/ravi
|
|
|
|
|
Hi,
I have an instance of data table named Bank_Details with the
columns "Bank_Code, Name, Acc_No",
sitting inside a dataset MyDataSet.
I have a BindingSource associated to the MyDataSet/Bank_Details.
Without creating another instance of the table, I want to bind
it to a datagridview, and only have Bank_Code column
show up in the DataGridView.
One way I know I can do it is to do a Visible=false on DataBindingComplete event for columns
that I don't want displayed.
But I am trying to avoid all that unnecessary adding of columns
and hiding them when my requirement is to not deal with those columns
at all.
Just wondering if there is a way to bind a DataGridView to an
already available instance of a DataTable and limit the databinding
to only specific columns?
I appreciate you all sharing any insights regarding this.
|
|
|
|
|
I've encountered this as well. If there is a solution that someone else has found that works better, by all means use it, but here's the workaround that I've used:
What I have done to get around this is to manually enter all the columns into the DataGridView in the forms designer, and set the "DataPropertyName" property on each column to the appropriate DataTable column that will be in the dataset. I can then set the visible on these to false, and bind to them normally, and have only the desired columns show up. What this also allows me to do is use the hidden values in rows for calculations and whatnot, but also set up formatting on the individual cells at design time instead of during runtime.
It works for my means (I use DataGrids to display a lot of line-driven real estate calculations and different views on the same value ($/SF/Mo vs $/SF/Yr in unbound columns based on read-only data in visible columns and other data in !visible columns).
Hope this helps
##EDIT
Edited to be a little more specific
|
|
|
|
|
Are you actually adding the columns one by one? Or are you just doing:
myDataGridView.DataSource = myDataTable
I haven't looked at it yet, but couldn't you actually just do:
myDataGridView.DataSource = myDataTable.Columns[x], to only show that one column?
|
|
|
|
|
I have a string that whenever it finds \r\n in the string i need it to be split, i tried using Text.Split("\r\n") but it doesnt like that. This is the only thing that is causing me errors with my printing, can anyone help me fix this? or does anyone know how i can count the number of times that \r\n appears in a string?
|
|
|
|
|
Right after i posted this i had an epiphany, i just replaced \r\n in the string with ` and split using that becasue i i was trying to find an odd character that no one uses very often and that solved my problem
|
|
|
|
|
You don't have to reserve a special character, just use text.Replace("\r", "").Split("\n")
|
|
|
|
|
Hello,
Instead, you should use:
Text.Split(System.Environment.NewLine);
All the best,
Martin
|
|
|
|
|
I do not agree.
First of all, System.Environment.NewLine results in a string, but String.Split requires
a char or char array.
Second, the logic you follow is correct only if the system that generates the file,
and the system that processes the file, both have the same opinion on how to
terminate lines.
e.g. if you generate the file on Linux, and want to process it on WinXP, it would
not work.
Luc Pattyn
|
|
|
|
|
I do agree with you in every point.
Luc Pattyn wrote: First of all, System.Environment.NewLine results in a string, but String.Split requires
a char or char array.
I shout test before I post!
Luc Pattyn wrote: Second, the logic you follow is correct only if the system that generates the file,
and the system that processes the file, both have the same opinion on how to
terminate lines.
It's not important in the application I'm working on, but it's good to inform about that it's possibility!
Thanks
All the best,
Martin
|
|
|
|
|
Is there a way of doing this? so i could have different registry on my memory stick.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
