Comments on: Multiple Degrees of Correctness http://blog.mozilla.com/tglek/2007/10/04/multiple-degrees-of-correctness/ Just another Blog.mozilla.com weblog Sun, 12 Feb 2012 07:38:09 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Hub http://blog.mozilla.com/tglek/2007/10/04/multiple-degrees-of-correctness/comment-page-1/#comment-4614 Hub Mon, 15 Oct 2007 18:32:35 +0000 http://blog.mozilla.com/tglek/2007/10/04/multiple-degrees-of-correctness/#comment-4614 re: the "monopoly", I think the first thing to do is chip heads at the CRTC, as the CRTC has proven to be innefficient and not protecting the public interest as it allowed the Fido-Rogers merger. On the other hand, Europe is very behind Canada in term of voice communication pricing as calling a cell-phone from a land-line still cost a lot (more than calling, say, north-America). And I see this on the pricing for international calls. France: 0.05$ / min, France (cellphone): 0.29$ / min. I guess there is not perfect world. re: the “monopoly”, I think the first thing to do is chip heads at the CRTC, as the CRTC has proven to be innefficient and not protecting the public interest as it allowed the Fido-Rogers merger.

On the other hand, Europe is very behind Canada in term of voice communication pricing as calling a cell-phone from a land-line still cost a lot (more than calling, say, north-America). And I see this on the pricing for international calls. France: 0.05$ / min, France (cellphone): 0.29$ / min.

I guess there is not perfect world.

]]> By: Jeff Walden http://blog.mozilla.com/tglek/2007/10/04/multiple-degrees-of-correctness/comment-page-1/#comment-4441 Jeff Walden Sat, 06 Oct 2007 01:23:55 +0000 http://blog.mozilla.com/tglek/2007/10/04/multiple-degrees-of-correctness/#comment-4441 In my mind the integer/boolean type collision just reinforces the need in C/C++ for a generative typing construct other than struct/class. I ran into this, hard, recently in an operating systems class distinguishing between kernel virtual addresses, linear addresses, and physical addresses, which we could only segregate into two different types (basically uintptr_t and void*) because otherwise we'd lose operator support. My final assignment suffered from three specific bugs, one of which would have been caught by having three disjoint types. I'm sure the typing distinction would also have sped up my rate of generation of the address-manipulation code, too. I think my next biggest complaint, for C, would be for a template system which could guarantee type safety without duplication by only allowing pointer types as type variables, but that's a rather different can of worms. I'm not sure whether it would be more or less useful than the previous complaint if implemented, tho. One of these days I'll take the time to look at the C-derived languages that add these features, but I'm unfortunately too busy at the moment to do so in any depth. In my mind the integer/boolean type collision just reinforces the need in C/C++ for a generative typing construct other than struct/class.

I ran into this, hard, recently in an operating systems class distinguishing between kernel virtual addresses, linear addresses, and physical addresses, which we could only segregate into two different types (basically uintptr_t and void*) because otherwise we’d lose operator support. My final assignment suffered from three specific bugs, one of which would have been caught by having three disjoint types. I’m sure the typing distinction would also have sped up my rate of generation of the address-manipulation code, too.

I think my next biggest complaint, for C, would be for a template system which could guarantee type safety without duplication by only allowing pointer types as type variables, but that’s a rather different can of worms. I’m not sure whether it would be more or less useful than the previous complaint if implemented, tho.

One of these days I’ll take the time to look at the C-derived languages that add these features, but I’m unfortunately too busy at the moment to do so in any depth.

]]>