MUDA の LLVM IR のバックエンドを書いていて float の即値のあつかいでハマりました. float の値によってはアセンブラ(llvm-as)がエラーを出す. なぜかと思って、LLVM IR のドキュメント(LangLef.html)を見たことろ、 Floating point constants use standard decimal notation (e.g. 123.421), exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal notation (see below). The assembler requires the exact decimal value of a floating-point constant. For example, the assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating [...]
CentOS 5.2/RHEL 5.2 comes with a very highly modified Xen 3.03 which if I'm correct is in fact Xen 3.1 backported. If you wan to use the latest Xen 3.2.1 you need to update the hypervisor. This tutorial is for x86_64 because that's what I'm running on grantmcwilliams.com. If you want the 32 bit version modify the appropriate paths and file names.
Wired: "Internet service provider Embarq eavesdropped on the web surfing habits of 26,000 customers in Kansas without notifying them personally, as part of its test of new, controversial advertising technology that profiles users, the company told federal lawmakers Wednesday."
Face of Daniel Molkentin (danimo)Inspired by the graphics used in this blog entry, I was wondering how hard it would be to actually make some buttons and banners on my own. Turns out it wasn't so hard after all. Enter the new Graphics page. Please make use of it!

I'll make a start:

Since I'm an IRC addict and I always liked to spend time learning new stuff on IRC, here is a review of several IRC clients for Linux. Some of them are well-known and popular, like XChat, Konversation or Irssi, and others are not so widely used or known, but nevertheless, they deserve mentioning.

I’m sorry that this has taken so long to post, but here is, at last, the promised summary of the answers I received in response to my question about licences. If any of my attribution links are wrong, then please do email me so that I can fix them. I have summarised people’s responses unless they were short, however I am using people’s names; if you wished to remain anonymous then I’m sorry, and if you email me I’ll fix the posting.

Joe Buck provided me with a well worded response which essentially boiled down to ”Pete is likely to be violating Fred‘s copyright” and “even if legal, Pete‘s action is not terribly respectful of Fred”.

James Ogden suggested that providing the plugin interface to FantasticUsefulApp, against which IncrediblyCoolPlugin was written, was both public and able to be used by a hypothetical FantasticallyUsefulApp++ written by another third party such that IncrediblyCoolPlugin was immediately usable without changes (or recompilation) then there is likely to be no issue with the GPL vs. proprietary part of the question. As for bundling it all together though, James couldn’t quite decide if this meant that SuperMegaCorp had run the risk of violating the GPL or infecting their app.

Craig Sanders gave a detailed response which I will attempt to summarise as follows: The answer to 1. is that the user is free to do as they wish wrt. linking on their own machine. No violation has occurred. 2. is a violation because the bundling of the plugin is not mere-aggregation in the terms of the GPL. 3. is not necessarily a violation of the GPL but definitely feels unethical. Craig also points out that 3 is an interesting turn-around of how Debian installs the Microsoft core fonts, the flash plugin, etc.

Neil Williams presented a strong argument that the BSD licencing of the header files was a red herring and that what matters is the licence of the object code implementing the interface, which clearly is proprietary in this example. He again argued that 1. was not an issue as there was no distribution of the object code involved. That 2. is a clear violation of the GPLv2 on the part of SuperMegaCorp and that 3. was a violation of the spirit of the GPLv2 even if it were strictly speaking within the terms of the licence.

Then the FTF-Legal mailing list was given a crack at the question. Unfortunately I can’t find a link to the thread, and I doubt this summary will do it full justice, so I guess someone needs to email me with a link to the thread. Initially the impression was that the only issue would arise in legal terms, were the plugin to be distributed under the terms of the GPLv3. This was countered with the comment that if age were the only factor in determining derivative works, then once SuperMegaCorp released a new version of FantasticUsefulApp it could be considered a derivative work of IncrediblyCoolPlugin despite that not necessarily being the case. Also the question arose about whether 2. could be considered more than mere aggregation if the app were advertised as having the features of the plugin. Finally there was the suggestion that Pete could simply allow for linking to the app in his licencing of his plugin, however I’m not sure if that’s possible given he aggregates Fred‘s work in his plugin.

Ean Schuessler provided a pithy response which given its terseness I shall include here in full:

Pete violated the GPLv2 when he introduced dependencies on FantasticUsefulApp into a derived work of HandyStuffs and subsequently distributed it. If Pete had kept IncrediblyUsefulPlugin to himself then he would have been fine. Uhura is also, arguably, guilty of at least contributory infringement by participating in the distribution process.

Several others, some of which wished to remain anonymous, also replied through various means which were not email as I had asked. I’m afraid I didn’t spot blog postings made in response, so if any of you did, and feel it can add to this redux, then please email me so I can add a link. Everything else was essentially the same as, or equivalent to, all the responses I have listed above.

Thank you all for your time and effort in this, the information gathered has been very interesting and will definitely shape how I proceed with writing plugins, and apps which use them, where any non-free licence may get involved.

About this series This series of three articles looks at PS3 Linux as a prospective development environment.This first article, Part 1, introduces the basic configuration knobs and widgets specific to the PS3, shows you how to use them effectively, and suggests the kind of trickery that might get improved performance or a more usable display. [...]
If you get the following with toped 0.9.0
Toped can't obtain required GLX Visual. Check your video driver/setup please

This is because a graphic card driver is not installed.
Mine is :
.Section "Device"
. .Identifier "Videocard0"
. .Driver "radeon"
.EndSection

That's the Xorg driver. It doesn't seem to maintain accumulation buffers and upstream needs them for the interactive graphics. On fedora 8, there should be no problem.

Upstream is trying to get to Xorg people about this, since weeks. This is the trouble from the very beginning (with f9). I didn't have a protection. That's another long story, but wx guys did something already, however it is not in their latest release. Whatever - the protection is now there, but it is not making the toped running for f9. On f9, there is no ATi or nVidia driver because of the state of Xorg and Mesa. We is trying to find a workaround because as it seems it will be quite some time until f9 gets vendor drivers.

The root problem appears when you run toped using GPL drivers - non vendor drivers. There is no protection and the program is giving segmentation fault. We have now patched toped 0.9.0. So if you run patched version (even on f8) without ATI or nVidia driver - it will not crash. Thus I'll recommend you to update your 'toped' package coming with the updates very soon.
Phoronix reports on developments with ATI Radeon R500 support. "This kernel mode-setting support is currently being housed in separate code branches, but will be merged to master and enter the Linux kernel in the future. David Airlie mentioned in the Radeon IRC channel that the Radeon KMS support will hopefully be merged right behind the Intel KMS support, but before that of course will be GEM going to master. David also hops to ship this ATI support with the Fedora 10 Beta."
Here's a quick way to report po-debconf transalations:



Courtesy of Quicktext


Update: the image above is an animated gif and some viewers might not be able to render such images correctly. Apparently even on the blog the animation isn't visible unless you click on the image itself.