It's true! ;) Apple is planning it's third grand transition (after the switch from the 68k to PowerPC and from OS9 to OSX). They plan to switch from the PowerPC to the Intel architecture in a period of two years. We should start seeing Intel-based Mac's as early as next year. I personally am a bit sad, I liked the PowerPC architecture and I hope IBM will pump some life to the PPC roadmap thus make Apple at least consider supporting multiple architectures. Hell, why not see OSX on the Cell processor? They have compilers that can compile to the both platforms and even make one binary that contains code for both platforms (they call them Universal binaries). But aren't there more that just two platforms? There is the PowerPC,the x86 (with MMX but no SSE, with SSE but no SSE2), the x86-64 (AMD wasn't even mentioned) and god knows what else you can pump into this list with time so you can quickly get a complex matrix you'll have hard work to keep small. Apple has to get really careful really fast as it's entering a domain where up until now MS has ruled unequalled. So the question is to they want to be the Apple that controls the whole stack from iron to OS or could they suffice with supporting OSX and selling also hardware of their own?
And don't think that MS will sleep over this as nothing has happened (on the other hand they do mention AMD quite a bit).
Anyhow, Jobs did an excellent keynote as always. There are quite a few comic moments (no spoilers here except the title :) also. I did feel that sometimes I saw glitches like the image-looking app that somehow seemed smoother on the Gx from a year ago. Maybe it's not optimized yet or perhaps the RISC architecture is better in handling large quantities of data then the CISC (or better, Intel takes the CISC code and translates it internally to RISC). Who knows.

Comments

Post a Comment

Popular posts from this blog

Relaxing SSL validation for JaxWS

I've recently had the need to access a web service developed in .Net and signed with a self signed certificate. I've decided to use the JaxWS and the Metro stack to develop the client and run it on Java6. As a plus, the service was protected with username and password. The service was exposed on an IP address and I repetedly had problems in establishing a connection. In the end, thanks to this article, it was obvious that the certificate was not created with the alternative name attribute but it was not an option to change the certificate as the web service was already used by other consumers (.net clients don't suffer by this issue). So, on my quest to relax the validation, I've found out this article and code snippet, which did not compile at first (I guess package names were changed in JDK6) so I've did some trivial refactoring and now, after invoking the static methods in the client code, the SSL connection gets validated with no problems. Hope it helps and th
Read more

Search and Replace in ODT using AODL

Recently I needed to create a way to search and replace a word in a OpenOffice OpenDocument ODT. It would seem the project itself already has a library AODL (take care, at least two other project pages but with older code) that exposes the format model but I didn't find any utility methods like search and replace functionality. I've created two versions of the method, one a old-style, imperative approach, and the other, a LINQ version. As I'm still new to LINQ it would seem that the classical approach produces a more efficient code. Anyone care to chip in and make a faster algorithm? In this particular version I'm just going for simple string equality, I didn't want to go with other StringComparison methods (just look at the Equals overload method). public static void SearchAndReplaceString(TextDocument document, string searchText, string replaceText) { var content = document.Content; foreach (var item in content)
Read more

Kaspersky Anti Virus 6 and Subversion problems

I've spent a good part of my day today in trying to understand why I can't checkout our subversion repository. I was getting this: svn: REPORT of '/svn/repos/!svn/vcc/default': 400 Bad Request google searches would make you think it's a proxy issue but I don't use or need one so after a lot of trial and errors I've found out that if you use Kaspersky Anti-Virus 6 it seems that if you use the Web Anti-Virus options with the recommended settings and you have the "Scan HTTP traffic" option on you end up with the problem above. It seems Kaspersky doesn't like the HTTP packets subversion uses so they get blocked. I really like Kaspersky and I recommend it to all but I hope they will address this issue as soon as possible as subversion is becoming the main open source VCS system. Any tips on this issue are welcome.
Read more