Welcome to Hardcode Sign in | Join | Help

May 2007 - Posts

Did you know that try { } catch(...) catching Access Violation exceptions is actually a bug ? Starting from Visual C++ 2005 it doesn't catch windows-based exceptions no more, only C++ exceptions.

Wow ! The movie has just won the top prize at this Festival de Cannes 2007 - this is absolutely incredible, I'm very happy and can't wait to see it !

 

If you want to deploy a database as MDF along with your setup, there's a big problem you will get with: where you copy the files so that SQL Server is able to read/write them ?

One method is to copy them in SQL Server's data folder, something like C:\Program Files\Microsoft SQL Server\MSSQL\Data. However, under Vista that folder is a protected area and, nevermind that, it's not very nice to copy files in SQL Server's folders.

The best method is thus to copy the MDF to your program's folder and then give read/write rights to the user SQL Server is running under. You can use EnumServicesStatus to find the SQL Server service, then QueryServiceConfig to find the user account and then the AddAccessRights from here to set the actual permissions.

Adrian points to me an excerpt from Windows Server 2008 product overview:

While the Server Core installation option is a fully functioning mode of the operating system supporting one of the designate roles, it does not include the server graphic user interface (GUI). Because Server Core installations include only what is required for the designated roles, a Server Core installation will typically require less maintenance and fewer updates as there are fewer components to manage.

I can bet remote access to manage the services will be possible via MMC, however - there will be no UI on the server or will we have a console UI similar to Linux or what ?

Yes it's official !

 

Did you know that 230 bytes is not called a gigabyte but rather a gibibyte ? WTF ?

 

Via feeder

I haven't participated as I somewhat misunderstood the rules - I thought the code should be realistic and not implement a calculator the weirdest way possible, however this is interesting also :). You can inspect some pre-release code from the participants here.
The site has been down for like 12 hours, there are no explanations on their blog. I really hope they were not hacked or having backup problems.

Just seen some performance numbers under NDA from ATI's upcoming Radeon HD 2900 (R600 to be more precise), it's terrible and I'm very upset being a hardcore fan of them :(. The performance is on par with 8800 GTS and thus less than nVidia's top of the range 8800GTX.

If they really put only 16 texture units, shame on them, it's incredible to have a 512-bit bus, double of everything else, and not throw at least 32 texture units :(. 

 

ATI has been losing the edge since X800 when they did the mistake of not covering Shader Model 3.0. R520 and R580 were late and not so powerful, and now the R600, even more late and less powerful :(.

Yet another strange problem on nVidia hardware, this time specifically related to a GeForce 7500 LE. Whenever I rendered into a pbuffer large enough, glFlush and glFinish had no effect and the image came out drawn partially. I've tried even drawing something after the scene at a specific pixel location and then looping until glReadPixels read the correct value :D still the scene was not rendered completely. The only viable workaround seemed to be tiled rendering on lower resolution pbuffers (I've tried with 512x512).

The bug didn't appear with a GeForce 6600 however I left the tiled rendering enabled for all nVidia cards just to be sure :(. 

Ok so I got over this strange bug, while creating a pbuffer it appears that wglChoosePixelFormatARB from Forceware drivers does not like some DCs, even if the DC passed there should be useless. It simply fails with GetLastError 0xc0070591. 0x591 appears to be ERROR_DC_NOT_FOUND.

So beware what DC you send, I finally sent one that has a OpenGL RC associated and it worked fine. 

As Firefox 3 and IE 8 will support them, let's take a quick look at what they are meant to be. Basically they are a way of structuring data on the web using existing HTML tags or attributes. Some standards are proposed - like for events or for reviews. For example you will have to add class="hreview" to your HTML element where the review is written, enclose your rating with <span class="rating">, and so on.

Besides, the once abandoned semantic html tags (cite, legend, strong) are now encouraged to be used instead of only using spans, divs + css. Microformats can be used both for smart tag-like functionality inside the browser (like click on a recognized event and set an alarm for it) and moreover they will be useful to spiders in scraping data.

Whoa! Microsoft doing a cross-platform framework ? Supporting Python and Ruby ? Now really, what happened ? :)

By the way I'm pretty sure Silverlight will be the core technology for counter-attacking Google and online services.