Comments by YvesDaoust

YvesDaoust 7-Dec-18 12:02pm View

Why worry ? Your value is accurate to fifteen significant figures, the maximu that double precision provides.

YvesDaoust 18-May-16 15:08pm View

Sorry, no, I am busy.

YvesDaoust 18-May-16 9:44am View

Write a complete program. If you don't have the skills, ask someone around you.

YvesDaoust 18-May-16 9:32am View

Don't worry, this is normal.

YvesDaoust 9-Mar-15 16:44pm View

Read this: http://mathworld.wolfram.com/NormalEquation.html

YvesDaoust 2-Mar-15 2:47am View

Lookup http://en.wikipedia.org/wiki/Linear_least_squares_(mathematics)#Derivation_of_the_normal_equations

YvesDaoust 11-Aug-14 11:23am View

Did you try it ?

YvesDaoust 11-Aug-14 4:27am View

Thanks for the hints. Capturing the WM_PASTE message raises an annoying complication: as you need to override the WndProc for the control, you need to create a custom control just for that purpose. (Simple derivation of the TextBox class can't be used as it causes the Designer to stop recognizing the code, making the Form impossible to graphically edit :( )

YvesDaoust 11-Aug-14 2:37am View

What idea ? Using KeyPress was already described in the question.

YvesDaoust 10-Aug-14 10:13am View

Its a Windows Forms TextBox, not WPF. My question is about handling the paste action, I am comfortable with the keyboard input.

YvesDaoust 8-Jul-14 16:28pm View

"all" the points is a lot, as there are infinitely many of them :)
To compute a single point along the arc, consider the unit vectors s = SE/||SE|| and r (as above), and form the sum C + s.cos(t) + r.sin(t), where t is some angle in the allowed range. (You can compute this range from the chord length and the radius.)

YvesDaoust 4-Jul-14 2:45am View

Glad to know :)

YvesDaoust 11-Jun-14 17:01pm View

A test case with five points is not appropriate. With so few points, virtually any matching is acceptable, as you assume that there are outliers. Such methods are typically use with hundreds to tenths of thousands of points with a significant part truly matching.

YvesDaoust 31-Mar-14 5:31am View

You do not specify if the first and last bounds are considered to be "between" as well.

YvesDaoust 28-Mar-14 5:38am View

No, not clearer. You didn't answer my question. There is no relation between depth and color, there is a relation between depth and lightness. You must describe what you want in more precise terms.

YvesDaoust 27-Mar-14 7:06am View

Still pretty unclear what you want. Are you referring to the gray shaded sphere (leftmost picture) or the colored "sheets" inscribed in a sphere (next picture) ? Such effects are much easier to achieve using a true 3D model than with a flat one.

YvesDaoust 19-Mar-14 3:19am View

Your question is not explicit enough. Please tell more about the effect you need to achieve and show samples.

YvesDaoust 14-Jan-14 9:00am View

This seems to be doable: http://msdn.microsoft.com/en-us/library/fh1h056h.aspx

YvesDaoust 14-Jan-14 6:31am View

This is Solution 1. All I had to change in my code was inserting the line
[assembly:ClassInterfaceAttribute(ClassInterfaceType::AutoDual)]; // This will let VB6 pre-load the interface définitions

But of course, all the class wrapping code needs to be there as well.

YvesDaoust 16-Dec-13 5:32am View

Reason for my vote of 1 \n Lacks generality and uses poor techniques.

YvesDaoust 10-Dec-13 10:28am View

BTW: users of this forum don't like minimalistic explanations together with lengthy code, where the reader has to do all the work

YvesDaoust 19-Oct-13 6:13am View

I have been looking for such an easy solution for months (no daily :) ), and it was in front of my eyes !

YvesDaoust 18-Oct-13 12:40pm View

Hurray, by searching better, I found the GetStdHandle call which does the trick, and now it works !

Thanks alot. You can enter your comment as a solution, I'll validate it.

YvesDaoust 18-Oct-13 12:35pm View

Interesting. I think I once had a look at the Console API and couldn't find a way to get the handle to the application console.

YvesDaoust 18-Oct-13 11:00am View

This IS an ANSI escape sequence. They don't work in a Command Prompt window. Thanks for the attempt...

YvesDaoust 12-Oct-13 10:13am View

More precisely, closing the numerator, before the slash

YvesDaoust 12-Oct-13 6:12am View

Five left and four right. Is that balanced ???

YvesDaoust 7-Oct-13 14:55pm View

I like your suggestion :)

YvesDaoust 7-Oct-13 12:26pm View

Extra hint about run-length-coding the shape: to fill an arbitrary polygon, the sweepline approach is well known (http://www.cs.princeton.edu/courses/archive/fall99/cs426/lectures/scan/sld028.htm). It amounts to decomposing the polygon into trapezia (some of which degenerate into triangles). You can adapt it to the case of a Pie, provided you take the curvature of the circle into account and split the arc at the top and bottom if needed to ensure that an horizontal line crosses once only.

YvesDaoust 7-Oct-13 12:13pm View

SetPixel/GetPixel are a severe bottleneck. In your case it is much better to "LockBits" the bitmap so that you can access the image buffer directly (you will find posts on this topic elsewhere in CP). If in addition you can generate a run-length-coded representation of the pie (not an easy task), you can memcpy whole runs at a time. This can be quite efficient for large areas.

YvesDaoust 7-Oct-13 11:39am View

How do you perform "per-pixel analysis" precisely ? There are ways to access the bitmap pixels efficiently.

YvesDaoust 7-Oct-13 11:22am View

I assume (not tested) that you can do without naming the shape:
With ActiveWindow.Selection.ShapeRange(1)
...
End With

YvesDaoust 7-Oct-13 10:57am View

Thanks for the suggestion

YvesDaoust 16-Sep-13 9:13am View

Reason for my vote of 3 \n Interesting

YvesDaoust 6-Aug-13 6:52am View

Yep, seems difficult to lower to 4 operations. Given the small domain, it is thinkable to fully tabulate the function (2^8 * 8 * 2 entries = 4 kB), but even the table lookup has a cost...

YvesDaoust 6-Aug-13 6:22am View

Nice try. The expressions are indeed correct. I'll remember the -b trick!

Unfortunately in my context I cannot precompute the mask m (and even less its complement), so that the operation count is not reduced.

YvesDaoust 19-Jun-13 8:47am View

Where is the problem ? You allocate 200,000 objects each 100 bytes in size don't you ?

YvesDaoust 17-Jun-13 15:14pm View

How exactly do you check the returned value ? This looks like an uninitialized value and I doubt this is the true returned value.

YvesDaoust 5-Jun-13 6:03am View

thanks for the feedback :)

YvesDaoust 4-Jun-13 13:35pm View

I didn't mean that you should redistribute the DLLs. I just meant that the redistributable are not preloaded, they have to be redistributed otherwise there would be no point making them redistributable. Use static linking.

YvesDaoust 4-Jun-13 11:27am View

Have a look at the Regional Settings on your machine (Control Panel). It can be that the thousands separator is set to be the comma. Defining it as empty could help.

YvesDaoust 4-Jun-13 11:22am View

I am not sure that the Windows operating system is deemed to have preinstalled run-time libraries (DLLs). They are not part of the distribution.
The run-time libraires are meant to be used by application developers with Visual Studio. Microsoft grants you the right to redistribute them. Windows utilities should not require them.
/MT and /MTd will link statically with the required run-times libraries and the executable will be self-contained. Anyway, mind possible incompatibilities with third-party libraries.

YvesDaoust 4-Jun-13 11:13am View

Are you sure you need to do that ? If passing 4 arguments instead of 1 is an issue for you, you can pass a, b anc c as global parameters (and keep calling the function F).

YvesDaoust 31-May-13 5:38am View

The code is working. I am confident that security rights and threading issues are correctly dealt with (even if these matters are totally and forever opaque to me) as all following operations succeed. The problem is silently ignored in a normal execution, it is just the debugger that tries to challenge me.

Thanks again.

YvesDaoust 31-May-13 4:13am View

Hi KarstenK,

thank you for taking the time to look at my question.

The good news: this sample program indeed works (provided you remove an extraneous pclsObj->Release() at the end, which causes an access violation exception (!)).

The bad news: the problem is still there.
It does not lie in the initialization function (I already tried three different options, to no avail: CoInitialize(0), CoInitializeEx(0, COINIT_APARTMENTTHREADED) and CoInitializeEx(0, COINIT_MULTITHREADED) - these three work with the sample program.)
I also tried the additional call to CoInitializeSecurity, which is missing in my code. This did change the behavior in some versions of my program but not in others.
So I conclude that the problem is memory-mapping dépendent or due to a race and will be harder to trace as it is somewhat random.
Anyway, your suggestion opened new ways to investigate, thanks.

YvesDaoust 21-May-13 9:24am View

This code does not even compile !

YvesDaoust 18-May-13 13:19pm View

It will be simple in any language (that supports floating-point arithmetic and 2D arrays), as you essentially need to encode the formulas for the 4 basic operations on complex numbers where you did it for floats.

YvesDaoust 18-May-13 13:12pm View

This question is unclear. What do you call a 2D spherical image ?

YvesDaoust 18-May-13 11:43am View

Your question is unclear as I (we) don't know what you call an "ordinary binary tree" vs. a "binary search tree".

YvesDaoust 4-Mar-13 12:37pm View

How do you submit the documents for printing ? Only this "print handler" knows how many pages were requested per document.

YvesDaoust 4-Mar-13 7:13am View

Thanks for the suggestion.

I have now completed a solution that does not require programming (well, a single line of code), and... works, so I am adopting it :)

YvesDaoust 26-Feb-13 6:18am View

Next trap, next solution: it is not enough to raise the ComVisible attribute, as no class method will be exposed (the class data members are). Now we need to let the compiler generate the proper interface definitions with ClassInterfaceAttribute(ClassInterfaceType::AutoDual).

YvesDaoust 26-Feb-13 6:11am View

Next progress: RegAsm seems to be case unsensitive as it raises an "ambiguous name" error in case two identifiers are identical except for casing. Once this is avoided, a TLB file is generated.

YvesDaoust 25-Feb-13 14:24pm View

Second progress: need to find the right version of regasm ! (I had several of them on my machine). Shame on MS for not warning about version incompatibilities.

YvesDaoust 25-Feb-13 12:44pm View

Do you need to find all N first digits, or just the Nth ?

YvesDaoust 25-Feb-13 12:42pm View

Your problem statement is incomplete. You first need to decide what method to use for the computation.
If the goal is just to practice parallel programming, you can pick any algorithm, even inefficient.
If the goal is to really accelerate the computation, then you need 1) to compute a huge number of decimals (otherwise it is not worth the trouble) and 2) to have a truly good sequential algorithm.

YvesDaoust 25-Feb-13 12:36pm View

What is Alchemi ?

YvesDaoust 25-Feb-13 12:30pm View

Dropping this is not an option, it is required by customers. The alternative is to rewrite the whole stuff as ActiveXs, which I want to avoid.

What do you mean by "contaminates the system" ?

YvesDaoust 25-Feb-13 8:07am View

As I already explained, there is no Build tab in the projects settings, I am using VC++ to build the assembly, not C# nor VB. I didn't see any option similar to "Register for COM Interop".

YvesDaoust 25-Feb-13 6:29am View

First progress: ComVisibleAttribute requires a "using namespace System::Runtime::InteropServices;" clause.

YvesDaoust 25-Feb-13 6:05am View

Just seen it and tried it. Things just a little more obscure now...

YvesDaoust 25-Feb-13 6:03am View

Something rather puzzling:
regasm mVizNET.dll /tlb says:

RegAsm : error RA0000 : Failed to load 'mVizNET.dll' because it is not a valid .NET assembly

I can't understand this, as the file mVizNET.dll does work; in particular, it can be referenced by a project and the Object Browser views it seamlessly. So this must be a valid assembly, mustn't it ?

YvesDaoust 25-Feb-13 5:46am View

Of course. None of it seems applicable to my case. Specifically: "5.Click Configuration Properties, and then click the Build node": there is no Build node in the configuration properties of VC2005; I coundn't find any equivalent (same about "Compile"); "Use the ComVisibleAttribute attribute": the syntax [CombisibleAttribute(true)] genrerates a compilation error; "The /tlb: switch generates and registers a type library": no, I don't see any TLB file generated.

This article makes me believe that what I want to achieve is possible and at the same time sticks me deeply.

YvesDaoust 25-Feb-13 2:33am View

Why the @ character ?

YvesDaoust 19-Feb-13 5:31am View

You should use the third argument to getline and specify the newline character as the string delimiter, otherwise space will be taken by default.

YvesDaoust 19-Feb-13 5:28am View

As far as I know, getline does not "ask" anything. It is an input-only function and does not emit any message. What makes you think the behavior differs when called from main() ?

YvesDaoust 19-Feb-13 5:21am View

How do you know it was skipped ?

YvesDaoust 19-Feb-13 2:16am View

Platform and language are of little relevance here, the SDK is multi-environment. The tag FreeImage was rejected. I am specifically adressing the members used to this package.

YvesDaoust 18-Feb-13 12:24pm View

A silly question: did you ever... call these functions or methods ?
And if yes, did you ever... pass some arguments ?

YvesDaoust 4-Jan-13 9:47am View

It seems that my problem is not isolated: http://social.msdn.microsoft.com/Forums/en-US/vcgeneral/thread/b2aacd05-d9e9-4c6f-95f5-d6c005c9a5c6/
Thanks for your contributions.

YvesDaoust 3-Jan-13 11:44am View

That's what I thought too. That's how it used to be. The Win32 projects are running fine, both Release and Debug. As I said, any x64 project fails in the Debug mode, even when created from scratch and even after complete reinstallation of VS2008 !

YvesDaoust 3-Jan-13 11:30am View

I don't plan to install on another machine. I just want to debug on mine.

YvesDaoust 3-Jan-13 11:13am View

Unfortunately, the target storage may not be an NTFS disk. I'd prefer a solution such that encryption/decryption is done locally before/after the transfer.

YvesDaoust 3-Jan-13 10:17am View

I am not sure you understood my post.

YvesDaoust 3-Jan-13 10:15am View

I am not shipping anything, just debugging a program.
I find it hilarious that you need to build an installer as a workaround.

YvesDaoust 27-Dec-12 6:25am View

Ooops, wrong, after a full rebuild it seems to work.

YvesDaoust 27-Dec-12 5:56am View

I don't believe that the problem lies directly in my code. It occurs in an early stage of application loading. Only trivial constructors are called for a few static class instances (mere intialization of data members, if initialization at all). No heap allocation on my side (no malloc, no new, no pointer dereference).

YvesDaoust 27-Dec-12 5:47am View

The situation has worsened a little: I have moved my global variables as static inside functions (that was relevant in my case). Now the Debug version does not crash anymore and the Release one still crashes, and does never trap into the debugger. So there is even less that I can investigate.

YvesDaoust 25-Dec-12 4:50am View

I wonder what extra information I could provide.

I tried to reproduce this in a toy project, to no avail. I have other application projects using the same library, that seem to work fine. The application that crashes is a much bigger one, but identical settings are used.

YvesDaoust 25-Dec-12 4:34am View

Yes, this is frequent. Impossible to trace as this is in system code and not trapped by the debugger.

YvesDaoust 17-Dec-12 7:57am View

Reason for my vote of 2 \n See comment

YvesDaoust 10-Dec-12 3:47am View

The answer is platform (OS) dependent. You tell us nothing about it.

YvesDaoust 4-Dec-12 15:37pm View

You're welcome

YvesDaoust 4-Dec-12 15:37pm View

High 5

YvesDaoust 4-Dec-12 3:31am View

The answer must be in the error messages. Read them carefully.

YvesDaoust 3-Dec-12 12:41pm View

Reason for my vote of 3 \n Make things as simple as possible, but not simpler.

YvesDaoust 3-Dec-12 9:10am View

In what way does this contradict my suggestion ???

1 and 3 can be done with a single variable; 2 requires two of them.

YvesDaoust 3-Dec-12 9:09am View

Deleted

In what way does this contradict my suggestion ???

1 and 3 can be done with a single variable; 2 requires two of them.

YvesDaoust 30-Nov-12 10:09am View

As I sais, implement a Pain handler. You can draw anything you like in it. And use the method Refresh of the form when you want to update the display.

As you are not sepcifying a freehand curve, but rather a polygon, you'd better implement a MouseDown handler.

YvesDaoust 30-Nov-12 9:36am View

I don't know how it is in ImageJ.

Capturing the cursor positions is very simple. Just implement a MouseMove event handler and store the coordinates in an array. Implement drawing in the Pain event handler. There is no need for a third party library.

YvesDaoust 30-Nov-12 5:26am View

Yep, you need to keep a count of words in the current sentence, and a count of words in the current paragraph. Can be done with a single map or two of them.

Storing the text structure is of no help.

YvesDaoust 30-Nov-12 4:33am View

I don't quite agree with this approach. It is overkill.

There is no need to store any hierarchical representation of paragraphs and sentences, as only counts are requested. Processing can very well be done on the fly.

The only relevant data structure I see here is a histogram of word counts, which is indeed appropriately implemented using a map.

Don't put classes everywhere. KISS.

YvesDaoust 30-Nov-12 4:16am View

You should specify whether the contour needs to follow some object edge/outline in the image or just be a freehand path.

First problem is hard; second is trivial.

YvesDaoust 22-Nov-12 6:46am View

I confirm that using FreeImage is not a seamless experience.

One of the failures of the test program is a source file not included in the project ! You wonder if the package was tested before release.

Another is a runtime error message you get when trying the same test program and using the test images: "*** TIFF Format
Error while opening TIFF: data is invalid ***"

Yet another, more annoying issue is the assertion error at run-time when running under Linux: Assertion `sizeof(uint64)==8' failed.

I guess I'll find my way through, but this is exactly the kind of time-wasters I wanted to avoid...

YvesDaoust 21-Nov-12 2:24am View

I tried FreeImage and it didn't compile under Windows unless I fixed the code. Then I tried a test program and it didn't link because of a missing function. Though I was trying the Windows-ready package for Visual Studio.

How can I trust a package in such conditions ?

YvesDaoust 16-Nov-12 10:44am View

yes, I mean no compilation issues, no need to fetch extra headers files, adjust linker settings, read endless manual...

YvesDaoust 16-Nov-12 9:43am View

Thanks.

Will it integrate seamlessly ? I think it is overfeatured compared to my needs.

YvesDaoust 4-Nov-12 15:16pm View

Don't go too far.

YvesDaoust 4-Nov-12 12:25pm View

Missing * operator.

There are also several errors related to coordinates computation. You made it more complicated that need be. Follow my guidelines.

YvesDaoust 4-Nov-12 8:26am View

Nice.

YvesDaoust 3-Nov-12 18:56pm View

Are you kidding ? This code already does a decreasing number of swaps per pass.

YvesDaoust 3-Nov-12 9:48am View

Nobody can help you as this question is perfectly unclear. Please rephrase.

YvesDaoust 2-Nov-12 9:45am View

You are right. I was referring to the Windows Forms combo. In WPF, you will use property comboBox1.TextBox.Text and event comboBox1.TextBox.TextChanged

YvesDaoust 2-Nov-12 9:30am View

You are welcome. You can rate my post then.

YvesDaoust 2-Nov-12 6:13am View

What is the real issue then, that you can't raise the event or that you can't get the changed content ?

YvesDaoust 2-Nov-12 5:52am View

Doesn't it work if you just assign a value to the Text property and handle the TextChanged event of the Combo ?

YvesDaoust 2-Nov-12 5:41am View

Your question is confusing. Do you want to disable clipboard operation for the Windows GUI (Windows Explorer for instance) or for your application window(s) ? In the second case, what kind of widget ?

YvesDaoust 31-Oct-12 12:55pm View

I want Visual Studio. My productivity with Eclipse is much lower, this is why I want to bypass it. It is slow and error prone. Make is not in order.

YvesDaoust 31-Oct-12 11:49am View

I must compile with all my compilers (6 or 7 of them) to check portability. This is my daily preoccupation. I'd like to simplify my configuration by avoiding Eclipse.
It indeed seems that VC6 was able to export makefiles. But I don't see this feature in later versions :( And I wonder if that would fit with my requirement #1.

YvesDaoust 31-Oct-12 11:16am View

Because I need my code to be compliant with another compiler.

How do you turn a VS project into a makefile ??

YvesDaoust 26-Oct-12 7:07am View

Yes, much longer. Bad for the caches and for virtual memory. It increases the amount of space that the processor needs to keep close.

YvesDaoust 26-Oct-12 7:01am View

The issue is that when going column by column the addresses go back and forth, and the successive pixels you visit are far apart.

YvesDaoust 26-Oct-12 6:56am View

What you wrote is 95% valid Visual Basic. Just enter it and fix the syntax errors. (I am not going to do that for you.)

YvesDaoust 26-Oct-12 6:53am View

As said, to work by increasing addresses. The column-by-column scheme stresses memory management too much, it lacks locality.

YvesDaoust 26-Oct-12 6:26am View

This is becoming quite surrealistic.

Do you realize that you gave the solution at the beginning of your post?

YvesDaoust 26-Oct-12 6:18am View

Yes, this is clear from the beginning.

You already know how to display a message box and input numbers, don't you ?
You know how to respond to GUI events like a button click, don't you ?
You also know how to perform a linear search in an array, don't you ?

So what is missing ? Just how to compare two numbers ?

YvesDaoust 26-Oct-12 6:06am View

Not sure if it is a good idea to carry out a task that has not been specified to you.

Anyway, the first question to answer is whether the categories will be determined by a human (supervised clustering) or established from the dataset by automated means (unsupervised clustering.)

YvesDaoust 26-Oct-12 5:59am View

I am doing my best to figure out what you can't do. You give me little clue. Please explain.

YvesDaoust 26-Oct-12 5:56am View

You are welcome.

There are several options for mapping complex numbers to RGB (2D unbounded plane to unit cube). One is Phase -> Hue, Magnitude -> Lightness (rescaled), fully saturated colors. It is an interesting challenge to make such transforms fast.

YvesDaoust 26-Oct-12 3:59am View

" that code is die " ???
" how to do " ???

Please rephrase.

YvesDaoust 26-Oct-12 3:38am View

This is a really unexpected statement, as 1) you understand the concepts, 2) you are quite able to use the VB IDE, 3) you wrote the code for a linear search, and yet you claim that you cannot do it !?

We always do things we never did before. What are you missing ?

YvesDaoust 26-Oct-12 2:33am View

So you've got all ingredients now, just code the search routine...

YvesDaoust 25-Oct-12 14:16pm View

You are welcome.

Just copy/paste the code and do the required adaptations to turn it to valid Java.

YvesDaoust 25-Oct-12 8:41am View

You should have stated your question differently then.

Do you have a Visual Basic development environment ?

YvesDaoust 10-Oct-12 6:09am View

thumbs up

The next battle is to lobby some standardization commitee so that they define tags for use on form so that your personal data can be filled in automatically...

YvesDaoust 10-Oct-12 5:59am View

I just tried it and it seems to be doing exactly what I want ! Thanks !

YvesDaoust 10-Oct-12 5:58am View

Thanks but I am not programming an application.

YvesDaoust 2-Oct-12 4:57am View

Please tell us what you mean by "best".

YvesDaoust 24-Sep-12 14:53pm View

Reason for my vote of 3
Usually, IRR computation is done using Newton iterations rather than dichotomy. Faster.

YvesDaoust 21-Sep-12 4:40am View

Have a look at those new errors and fix them. In the first place, include <windows.h>.

No I am not trying that for you.

YvesDaoust 14-Sep-12 14:06pm View

I made yet another attempt with the FilgraphManagerClass object. It has all the bells and whistles I need, like RenderFile, Run, Pause, GetCurrentImage, CurrentPosition...

Unfortunately, I found no way of getting frame-by-frame resolution :(

YvesDaoust 14-Sep-12 3:08am View

I made an attempt with FFmpeg and quickly ran into portability problems with the MSVC environment (missing include files, vararg macros...), all requiring ad-hoc adaptations at various places.

This is exactly what I want to avoid. Dealing with such mammuth packages is a pain in the neck.

YvesDaoust 14-Sep-12 2:25am View

No I didn't, thank you for pointing me at it.

Unfortunately, I tried the IMediaDet interface and it does not work with my file. It just reports 0 output streams in it.

This is confirmed by the Demo application that comes with the article.

YvesDaoust 14-Sep-12 2:20am View

I don't mind mixing native and managed code. I am looking for a solution that I can put at work in a couple of hours. Can you recommend me one that has a straightforward API?

YvesDaoust 5-Sep-12 4:18am View

Interesting, thank you for contributing. Where did you find this info ?

This is perfectly horrible, I will stick to my Solution 1 !

By the way I am nearly done rewriting all my stuff using intrinsics. In the beginning I was strongly against them, considering that they were just a painful rewrite with no added value. After a while I realized that they nicely take care of register allocation and make your code look human again. I saw no loss of performance compared to my carefully handcrafted assembly; sometimes improvements by 20% or so...

YvesDaoust 3-Sep-12 6:19am View

For me it is unclear what you cannot do.

YvesDaoust 3-Sep-12 5:54am View

What is the question ?

YvesDaoust 20-Aug-12 4:35am View

It is unclear if you want a tool that modifies the image (inlay) or keeps a set of drawing primitives on top of the image while keeping the latter untouched (overlay).

Also unclear what you mean by "redo some operation".

YvesDaoust 18-Aug-12 8:02am View

Thanks. I have tried sending to one of the bloggers. We'll see...

YvesDaoust 18-Aug-12 7:27am View

Microsoft's sites are a labyrinth. I need a way to address the right guy in the right team, or their "suggestion box".

I don't want them to solve this issue for myself (I have workarounds and I won't be waiting for the next VS release). I just want to make this suggestion with some probability that it is heard.

I am not sure it requires changing the compiler, rather updating the relevant table somewhere.

YvesDaoust 18-Aug-12 5:51am View

These are of no help. Why did you answer that ?

YvesDaoust 18-Aug-12 4:29am View

What do you mean ?

YvesDaoust 6-Aug-12 15:53pm View

In the end your suggestion is the right one. One option would be to retrieve the local variables from the stack frame, but it appears very difficult to locate them in optimized code. Passing them as argument should make things easier.

YvesDaoust 4-Aug-12 8:38am View

If you want to work with larger files, all versions of Visual Studio Express have a wonderful text editor that can load and process them.

YvesDaoust 3-Aug-12 4:31am View

From this quite interesting CodeProject article "Inline Assembly in GCC Vs VC", I understand that my problem is not specific to the Intel syntax: gcc inline assembly just does not support references to local variables. These have to be found in the stack frame :-( (hoping that they have not been optimized away...)

YvesDaoust 3-Aug-12 2:49am View

Not the right day to quit smoking, I am afraid.

YvesDaoust 2-Aug-12 15:26pm View

Thank you for trying to help.

This inline assembly is MMX/SSE code hand crafted for maximum performance. I don't believe that any vectorizing compiler could beat it.

There are more than 10,000 lines of assembly code so rewriting to AT&T syntax is not practical (not talking of the disgust it inspires me.) There is a customer waiting at the door, and the work is 99% done, except for this regrettable restriction.

I am dreaming of a better option than locating the variables on the stack for hundredths of functions.

YvesDaoust 2-Aug-12 10:54am View

What do you mean ? I am using functions.

My question is very specific to the behavior of gcc/g++ in the Intel inline assembly compatibility mode.

YvesDaoust 20-Jun-12 10:13am View

In C#, I work like this:

in Form1.cs, declare a class like this:

// Trick to allow double buffering
public class DoubleBufferPanel : Panel
{
public DoubleBufferPanel()
{
SetStyle(ControlStyles.AllPaintingInWmPaint, true);
SetStyle(ControlStyles.OptimizedDoubleBuffer, true);
}
}

Then in Form1.Designer.cs, replace the occurrences of the Panel type by the this newly defined class (two places).

YvesDaoust 31-May-12 11:38am View

Interesting proposal.

I am trying that, it works on a property and a method argument ! Thanks !

YvesDaoust 31-May-12 10:30am View

The class library is written in C++. The buffer arguments are typed as unsigned char* (unmanaged). I am not writing the VB declarations myself, the Object Browser does.

YvesDaoust 18-May-12 10:15am View

How can I restart the application after the reboot to where it was interrupted, in case of a desktop?

YvesDaoust 17-May-12 6:14am View

That's also my feeling, thanks.

YvesDaoust 17-May-12 6:08am View

Nice, thanks.

The MSDN topic is the most interesting: http://technet.microsoft.com/en-us/library/cc180825.aspx

YvesDaoust 17-May-12 6:02am View

Cool, new leads !

Any idea about the availability/reliability of the info ?

YvesDaoust 17-May-12 6:01am View

Deleted

Cool, new leads !

Any idea about the availability/reliability of the info ? [see my comment on Solution 2]

YvesDaoust 16-May-12 11:08am View

Yes, quite right. For me the module is more the size of the cells, but this is just a matter of convention.

YvesDaoust 23-Apr-12 11:21am View

The replacement approach is sound but has a potential flaw: a character could be replaced more than once if it appears in several confusions.

For instance, assume the confusions p <-> ph and h <-> H; pasoli would turn to phasoli, then to pHasoli.

YvesDaoust 23-Apr-12 10:03am View

I don't believe one can easily make sense of this problem statement.

You use the term "count" (usually meaning number of), but you compare sums (adding the values together).

And what do you call "swapped numbers" ? All of the numbers are moved during the sorts.

YvesDaoust 13-Mar-12 3:34am View

Let's work to help developers, not make them feel stupid.

YvesDaoust 17-Feb-12 8:47am View

The second part of the condition does not make sense; it is the same as f(n) != g(n), I don't think this is intended.

YvesDaoust 7-Feb-12 2:56am View

Deleted

Why do you use these spanLeft and spanRight flags ?

YvesDaoust 2-Feb-12 8:06am View

1) I agree with that
2) with that too
3) the same regedit.exe is run in both cases (the file associations were left as default on both machines), this is what makes the case unintelligible
4) this is exactly what I would like to avoid, I don't want this redirection

YvesDaoust 2-Feb-12 6:52am View

I agree with that, but I still want to understand what causes the redirection on one machine and not on the other.

BTW: I haven't solved this.

YvesDaoust 31-Jan-12 2:29am View

Deleted

Reason for my vote of 5
New, handy tool; not extra luxury when you think about how painful leaks hunt can be.

YvesDaoust 30-Jan-12 2:30am View

Nikhil, I think you don't realize that your question is meaningless without providing mode context. Avoid doing this, as it easily irks the CodeProject members.

YvesDaoust 25-Jan-12 11:29am View

By the way, your title is misleading.

YvesDaoust 24-Jan-12 3:01am View

Deleted

Reason for my vote of 5
You were the only one to raise an interesting issue regarding this otherwise trivial problem

YvesDaoust 24-Jan-12 3:00am View

Deleted

Who said that the empty string is not a palindrome ? ;)

YvesDaoust 24-Jan-12 2:39am View

Deleted

Reason for my vote of 5
Good to know, I would never have imagined that such an option existed by myself

YvesDaoust 19-Jan-12 4:17am View

Deleted

Too glad that you liked these.

The given implementation IS iterative; the trick is that the log10 invocations correspond to B constant values that just need to be precomputed (log10(0.5)=-0.301030, log10(0.75)=-0.124939, log10(0.875)=-0.057992...).

You will find plenty of CORDIC implementations in various language, but it is up to you to mine the Web.

Have a look at http://www.voidware.com/cordic.htm, and also http://www.quinapalus.com/efunc.html

YvesDaoust 18-Jan-12 3:05am View

Deleted

You are welcome.

For the sake of readability, you can map an RGB structure on the byte triplets. See my Alternative 5.

YvesDaoust 17-Jan-12 15:21pm View

Deleted

DISCLAIMER: someone inserted spaces on the left of the assignment operators; I hate that.

YvesDaoust 17-Jan-12 15:14pm View

Deleted

next time i'll sta

YvesDaoust 27-Dec-11 3:32am View

Deleted

Reason for my vote of 1
This code smells. Combining a monitor with busy waiting loops is a nonsense. This code is too complicated to be verified. I would never trust it.

YvesDaoust 22-Dec-11 10:07am View

Deleted

What tags ?

YvesDaoust 20-Dec-11 3:14am View

3.If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Let's work to help developers, not make them feel stupid.

YvesDaoust 14-Dec-11 9:26am View

Deleted

I thought you liked Monthy Python's sense of humour.

YvesDaoust 14-Dec-11 4:34am View

No. What more do you know about these images ? [You seem to know how the middle one was obtained]
If yo don't tell me the context I cannot help you.

YvesDaoust 13-Dec-11 5:40am View

Actually, it is impossible to answer your question without knowing how you would define a good lineament detection. What would you really expect as a valid output of the algorithm. Subjectivity should be dealt with in the first place.

If I were to do it, I would first try a classical edge detector (Canny + non-maxima suppression + hysteresis thresholding), then filter on edges that go in pairs (this would be an ad-hoc algorithm to be created).

You can also get some inspiration from http://www.irisa.fr/vista/Papers/2000_mva_laptev.pdf, even though this is a fairly sophisticated solution.

YvesDaoust 13-Dec-11 4:45am View

The median filter does not produce a binary image by itself. Do you you mean binarization followed by binary median filter, or grayscale median filter followed by binarization, or something different ?

In my not-so-humble opinion, this algorithm is a poor lineament detector. Indeed it generates a lineament-like image, but that has little correspondence with the underlying image. Too many false positives and false negatives.

YvesDaoust 13-Dec-11 4:14am View

What is binarization by medians ? Never heard of that. Median filter ?

I understand the intent to find lineaments. Anyway it seems to me that the method is very uneffective. May lineaments are detected where there are none in the source image, and conversely, many salients lineaments are ignored.

YvesDaoust 13-Dec-11 2:32am View

Deleted

Reason for my vote of 1
Is this article adding any value ?

YvesDaoust 1-Dec-11 4:27am View

If you never programmed in your life, take a few minutes to download Visual Basic Express edition (free) and play with it a little.

Follow the "Creating Your First Visual Basic Program" tutorial. You'll quickly understand if this is for you or not.

YvesDaoust 1-Dec-11 4:10am View

You can try by yourself first. Never too late to ask an expert after.

You might even be pleased that the manufacturer provides what you need as a sample program.

YvesDaoust 1-Dec-11 2:58am View

Wonderful, this closes my question. Thank you for providing this precious info.

YvesDaoust 30-Nov-11 7:03am View

Did you read the question ? The question is about the formats to support, not about coding.

YvesDaoust 30-Nov-11 6:59am View

Did you read the question ? The question is about the formats to support, not about coding.

YvesDaoust 30-Nov-11 6:59am View

Deleted

Did you read the question ? The question is about the formats to support, not about coding.

YvesDaoust 28-Nov-11 9:15am View

In terms of readability, your 1) is the coolest. [Except for the necessary reversal trick.]

YvesDaoust 28-Nov-11 9:04am View

Quite right. I was misled both by the code brwoser and by the documentation, which only report overloads with 1 to 3 fields.

Thanks for the tip !

YvesDaoust 28-Nov-11 7:54am View

Just too lazy to use an array

YvesDaoust 28-Nov-11 6:40am View

By the way, nobody's perfect: when printing Unicode characters to a Windows console, two bad things happen:
- printing with wprintf does not use the Unicode character set (€ appears as ?);
- printing with _putws just stops writing when it meets a non-ascii character !

YvesDaoust 28-Nov-11 3:11am View

Thanks for the answer.

My question is more about the client applications: do they all recognize Unicode content ? (the classical MS applications do)

YvesDaoust 27-Nov-11 12:20pm View

Given that you are using the built-in function for text replacement, there is little chance that you could achieve better performance from a Word macro.

If you really experience a perfomance problem (huge text/massive number of replacements), doing the replacements using other tools can be an option.

YvesDaoust 25-Oct-11 5:01am View

Deleted

Reason for my vote of 1
Why make it simple when you can make it dangerously complicated ?
Those two flags should be mutually exclusive. This opens a "correctness hole" in the code. A single flag makes this property implicit.

YvesDaoust 25-Oct-11 3:51am View

Deleted

Actually, this is not a joke. How can you check that a multiply has overflown ?

YvesDaoust 25-Oct-11 3:45am View

Deleted

Just for the sake of being picky, how can you be sure that when overflow does occur the condition long.MaxValue - answer < answer holds ?

YvesDaoust 19-Oct-11 4:35am View

Deleted

Matthew, have a look at this site: http://www.luschny.de/math/factorial/FastFactorialFunctions.htm. It is a wonder.

YvesDaoust 19-Oct-11 4:31am View

Deleted

Shouldn't you add overflow detection ? (Doing it without the knowledge that overflow starts at 13 - otherwise that would be cheating.)

Comments by YvesDaoust (Top 200 by date)