Clik here to view.
re: This code would be a lot faster if it weren't for the synchronization
@Henning: Good luck making bulletproof deterministic tests for all possible race conditions. Very excellent as your particular hammer may be, this isn't the nail you're looking for.@Neil: Um......
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
Therac-25 used the "no locks" solution. On a radiation therapy machine used by hospitals. The results were what you'd expect - fatal. See en.wikipedia.org/.../Therac-25
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
@Anon. The therac 25 problem was due to bugs, not due to someone trying to optimise the program. Duh.
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
[Neglecting the hardware reductions due to cost or whatever optimisation] Sorry - didn't consider that at first.
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
we did that in one of our apps, stub out all the lockincrement/decrement in a IUnknown implementation every object used. but we didn't really need it, not for every object anyway. it was just one of...
View ArticleClik here to view.
re: Now that version 4 of the .NET Framework supports in-process side-by-side...
Cygwin is proposedly not a subsystem, it's based on Win32. That's why its fork() sucks so much, among other things. But hey, if you're using Cygwin for production, You're Asking For It™.
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
Danny Moules: When the kernel is running on just a single processor, nothing else can be running simultaneously, so it doesn't need locks. Of course it can still be interrupted by hardware, so it just...
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
@Gabe:This is incorrect.The uniprocessor kernel (which doesn't exist anymore, by the way), has the same implementation of Enter/LeaveCriticalSection. Its implementation of spinlocks, however, only does...
View ArticleClik here to view.
re: When the option becomes so second-nature you forget that it's an option
Who hasn't had a brain-fart when it comes to switches?
View ArticleClik here to view.
re: When the option becomes so second-nature you forget that it's an option
"unregistered table server"... makes me think of someone bringing me food at a restaurant along the southern US border.
View ArticleClik here to view.
re: When the option becomes so second-nature you forget that it's an option
Everybody is conditioned to run "!debugextensionQ -s", when really "!debugextensionQ -r -t" is closer to what most people want. I guess nobody ever ran "!debugextensionQ -?" to see all the cool...
View ArticleClik here to view.
re: Now that version 4 of the .NET Framework supports in-process side-by-side...
@John Doe: No choice. It's the only to get a working sshd on Windows.Short comment is short.
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
@Gabe: you are completely wrong. Consider one thread reading a value from storage, incrementing it, and writing it back, while another thread reads the same value, increments it, and writes it back....
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
> Thread A can be interrupted by timer expirations after its read but before its write. In the meantime, while Thread A is suspended, Thread B can read the (old) value (that has not been...
View ArticleClik here to view.
re: When the option becomes so second-nature you forget that it's an option
Well to be fair a lot of programmers have little idea how to install things. They have IT lackies for that especially if they've worked for large shops with a team for testing a team for dev, a team...
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
The first 32-bit CPU I ever programmed (well.. I had been programming 36 bit ones earlier) had a user level instruction SOLO, disabling all interrupts for up wto 255 clock cycles, or until a TUTTI...
View ArticleClik here to view.
re: This code would be a lot faster if it weren't for the synchronization
Little things we ignore in life turn out to be most important. Same way your blog post is talking about "Presence of Synchronization" being the culprit but in reality its "Bad Algorithm" that's causing...
View ArticleClik here to view.
re: When the option becomes so second-nature you forget that it's an option
This is how people use the tar utility also, thinking that "xzvf" is some sort of indecipherable magic incantation that must be performed.(see also https://xkcd.com/1168/)Just learn what the letters...
View ArticleClik here to view.
re: Display a custom thumbnail for your application (and while you're at it,...
"This is hardly "basic Win32 stuff". It's "obscure historical Win32 stuff." -Raymond"So fundamental messages like WM_MOUSEMOVE and WM_LBUTTONDOWN are obscure and historical?What other obscure and...
View ArticleClik here to view.
re: Display a custom thumbnail for your application (and while you're at it,...
Which message crackers should we use with WM_DWMSENDICONICTHUMBNAIL then?[It doesn't have one, but you knew that. Look, I already said it was a mistake. What do you want, a time machine to go back...
View Article