Try freessh.org and openssh.org.

That's the sense I've gotten. I have a sense that the Windows model treats the USB bus like the PCI bus with no way to talk directly to devices. Unfortunately, there's no device driver for this unit. My friend is writing an open-source support library for it and it's been trivial to support it under MacOSX and linux without writing a device driver. I guess windows just won't make it.

Together is written in Java too, so speed should not be the issue. The bigger problem with Poseidon is that its UML model is too strictly tied to the Java language, so it doesn't always fit C++ programming very smoothly. Also, CASE is useless without round-trips and Poseidon only supports this for Java.

A friend has a USB device that he wants to support under Windows without dealing with the DDK and writing a device driver. The device is a data logger and he wants to talk to it in the same way he would if it were on an RS-232 port or a TCP/IP socket. He has working code for this on MacOSX using Apple's IOKit and on Linux using libusb. Is there an equivalent API for accessing a USB port in this way under Windows?

Thanks to everyone who helped explain this. Thanks especially for the code fragments. Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

Duh! Thanks for setting me straight.

Can anyone tell me why this code

#include <map>

std::map<std::string, int> foo;

int test_stl(std::string key)
{
    return foo.find(key)->second;
}

Gives me about a page of STL compilation errors in Visual Studio .NET? Can anyone give me an example of working code that uses a hash_map or a map with an std::string key type? Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

How can I make a hash map with an std::string for a key? VC++ .NET barfs, complaining that it cannot convert from type 'std::string' to 'size_t'. Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

mbh@ito wrote: Also Subversion >could be< worth looking into Subversion is in the early alpha-testing stages. Would you trust your production code to a version-control system that's only in alpha? Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

It should be possible to get LCD panels that can be selectively darkened. As to light loss, you are looking at a maximum of 50% transmission when it is set to transparent. This is because LCDs work only with polarized light and to polarize ambient light, you have to throw out (i.e., absorb) the 50% with the wrong polarization. LCDs can also show chromatic behavior where the attenuation varies with wavelength. There are materials that can become selectively absorptive or reflective, but these are generally somewhat exotic and quite expensive. One question would be how fast you need to switch. If switching speed is not critical, I would think about something microfluidic, where you would flow liquid dye of different concentrations into different cells in the device. Switching speed would be excrucuatingly slow, but you could get arbitrary transmissions between 0% and 100% with good achromatic behavior and a much better contrast ratio than LCDs could provide. I don’t think nation-building missions are worthwhile. George W. Bush

Thanks. That's just what I needed. Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

Thanks, but as I wrote, my application does not use .NET. It's a native, unmanaged executable. All I seem to find are .NET profilers when what I want is something to profile native code.

I am writing a machine vision application that must process 5000 frames per second from a high-speed camera. I am writing in unmanaged C with Visual C++.NET. I have to find the bottlenecks in my code. As far as I can tell from the VS.NET documentation, the only performance analysis tools that ship with VS.NET are for profiling distributed web applications and are not suitable for fine-grained source-profiling of unmanaged code. Is there any way to profile the source code with Visual Studio.NET or do I have to slip back to Visual Studio 6.0? Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

Let me give you an example of an article I wrote[^] that includes GPL code: I contribute to a project that is GPL. You can like it or not, but I find CVS a useful program and I fixed some bugs in it that relate to some severe bugs in the way NTFS handles dates and times on files. I thought others would be interested in how I worked around the Microsoft bug, so I posted an article and illustrated it with a snippet from the CVS source. Because CVS is GPL and my code derives from the CVS code, I cannot remove GPL from the example code I posted unless I were to rewrite it completely. I was too lazy to do that, but still thought it would be useful for CP'ers. I would think that it would be useful for people to see HOW to work around the NTFS date/time bug and the code is sufficiently simple that it would be straightforward for others who want proprietary versions to roll their own interpretation of the ideas I present. If this article were pulled because of a ban on GPL, would you really find life easier without the explanation of how to work around the NTFS date/time bug? Jonathan Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

The ICM 2.0 documentation seems to indicate that ICM can work with any valid ICC color profile. These profiles can specify non-sRGB color spaces, such as CMYK and CIE LAB. Does anyone know if it's actually possible to use Windows color management to work in CIE LAB space or CMYK space? This is relevant because I am really interested in being able to print to my printer's complete color gamut, not only the part that overlaps with sRGB. (The default sRGB color space that windows uses is really lousy at representing saturated yellow, orange, and cyan colors that standard printers can represent very nicely).

Spent 3 hours(!) running "Repair installation" and still I see the same thing. I guess it's just one of those mysteries that I'll have to live with. If anyone has any thoughts about how to diagnose and resolve a problem like this, I would be grateful to hear about it, but my guess is that it's just an obscure bug in VS.NET. Fortunately it only slows me down but does not stop me from getting work done.

Thanks. I just have the one machine, so it's hard to tell what's universal and what's a quirk of my machine. I guess it's time to uninstall/reinstall and see if the problem goes away. Why couldn't Science, in the long run, serve As well as one's uncleared lunch-table or Mme X en Culottes de Matador?     James Merrill

Is anyone else having this problem with VS.NET IDE? I can't navigate my source code using Class View or the object Browser.

1. In VS.NET, create a new project. Choose Win32 Project. Call it Foo.
2. In the ClassView pane, double-click on "InitInstance". The editor window will open Foo.cpp and highlight the word "InitInstance" on the first line of the definition of InitInstance. So far so good.
3. In the editor window, type <Ctrl+Home> to move the caret to the top of the buffer.
4. Now double-click on any function or variable name in Class View. The caret in the editor window remains at the top of the buffer rather than moving to the definition of the symbol you chose in Class View.
5. The same sort of problem occurs if you right-click on a function in ClassView and select "Go to definition" or "Go to declaration".
6. If you enable Browse information and build the project, the same behaviour occurs for Go to definition or Go to declaration in the browser window.
7. Using the combo boxes on top of the editor window, you can navigate properly (choosing a function name does cause the editor to highlight the function definition and move the caret to the correct place.
8. Closing the IDE, deleting foo.ncb, and reopening the IDE does not resolve this problem.
9. Rebooting does not resolve this problem.

This reproduces 100% on my machine, at least for yesterday and today. Can others reproduce this behaviour? It doesn't seem to appear in the VS.NET bug list at http://support.microsoft.com/common/canned.aspx?r=d&H=Documented%20Bugs%20in%20Visual%20Studio%20.NET&LL=kbvsnetsearch&Sz=kbbug%20and%20not%20kbnetfssd[^], so I have submitted it (Thanks, Nick, for getting the bug-report page fixed). Can't I just get upset without having to think about what I'm really upset about?    Six Feet Under

Thanks for the link. Unfortunately, I tried submitting the bug on that page and got:

State=37000][Error=4060][Microsoft][ODBC SQL Server Driver][SQL Server]Cannot open database requested in login 'Feedback'. Login fails.

Resubmitting gets the same response. Anyway, since you know about it from this forum, I won't lose any sleep over it and will assume that you have ways of figuring out what to do with the report (mv /dev/nul ;) ). It has always puzzled me why some folks at Microsoft (certainly not you) seem to put so many barriers in the way of receiving feedback from users. I would have thought that putting information on bug reporting somewhere accessible, such as on the VS.NET home page, would provide useful information to the VS.NET team. You say yourself in the article to which you thoughtfully provided the link, that "how do I submit a bug report" is a FAQ, which leads me to wonder why a FAQ is not answered in a reasonably obvious place on the MSDN web, the KB, or the VS online help.

When I try search-and-replace text in certain C files with Visual Studio.NET, I get a message box with a red X, the title "Microsoft Development Environment" and the message "Catastrophic failure". The development environment DOES NOT crash and if I repeat the search-and-replace, it seems to work. Is anyone else seeing this? Any idea what's causing it or how to avoid it? I get no hits on the Knowledge base when I search the knowledge base on "Catastrophic failure visual studio". Also, I can't find anywhere in the Visual Studio Help or on MSDN about submitting a bug report (searching for "Bug report" "Visual studio" on MSDN returns no relevant hits) and there is nothing about submitting bug reports on the "Visual Studio," "Customer Support," or "Contact Microsoft" pages at MSDN. If anyone from Microsoft is reading this, are you interested in bug reports and if so, how does someone find out where to submit them?