this week in curl land

It has been another busy week, and this time people brought up a range of stupid and annoying bugs:

libcurl…but also included Patrick Monnerat’s cool internal re-arrange to use handler structs for the various protocols.

Upcoming dates to check for:

  • October 14 – Feature freeze for the 7.17.1 release (and I take off for China)
  • October 21 – I come back home from China (to catch up with a million emails)
  • Octobert 27-28 (something) – Probable release date for the 7.17.1 release, assuming that all serious bugs have been fixed by then.

As usual, post curl-related stuff on the mailing lists and not to me personally!

gcc 4.2.x woes

In Rockbox we’ve seen the gcc 4.2.x series introduce warnings on C code that really is perfectly fine, and also compiles warning-less on all other versions of gcc. (We use the 4.2.x series in Rockbox only for the simulator, built for running on host.)

In the linux-mips project, they’ve found gcc 4.2.2 to not even be able to build their kernel…It has been suggested that perhaps the best thing is to just skip to 4.3. Bugzilla entry.

4K Makes Lots of Data for Future Kids

Yeps, it seems people these days do “4K” (4096 x 3112) movies when they want to be on the bleeding edge of digital movie resolutions. That’s more than 6 times the number of pixels of full HD (1920 x 1080).

Dalsa 4K cameraJonathan Schwartz of course sees an excellent opportunity to tout ZFS in this world of really really huge data amounts, since as he puts it “the digital master for an average 4k film is roughly 9 Terabytes – and with working material included of course a lot more.

Now, I figure this of course is a perfect market for huge data storage and file systems that can deal with this, but my gosh this will stretch their backup systems to the limit – not to mention the problem with data longevity. How will this material and data live and be stored for future generations to be able to take advantage from it? I already have thoughts about this for my digital images and video snippets, but when the world is going towards insane data amounts and for almost every part of life, I can see how we in the future risk having less traces left from the past than what we have today from our past…

Also, if bluray/HD-DVD is suitable for HD content, what on earth will we need for 4K content?

Rockbox USB

USB type A plugUSB in Rockbox land was originally a matter of supporting the USB bridge chips in the first targets we supported (the Archos ones, the irivers and the iAudios).

Since the USB stacks moved into pure software in the used soc chips, Rockbox has (unfortunately) relied on the original firmwares (the so called OF) to provide USB support so that host computers can access the players.

One of the projects in the Google Summer of Code 2007 for Rockbox was to introduce a USB stack and offer native USB support for Rockbox, at least on the PortalPlayer-based targets. These targets were selected because the PortalPlayer chips have been found to have a USB set that is next to identical to the one used in the Freescale i.MX31 and that is fully documented online. Christian Gmeiner took this project to state where it partially works, but not enough to be actually useful to any Rockbox user. Christian’s code was largely based on USB code from the Linux kernel.

Now, long time Rockbox hacker Björn Stenberg enters the stage. Being one of the (original) core guys, he has a firm believe in KISS and as such he has started over on a brand new USB stack implementation that is meant to replace Christian’s and to be smaller, less complicated and quite possibly end up actually working! Björn once wrote the ISD200 support in the USB stack for Linux, so he has been in this neighborhood before…

Segmentation Fault Fix

I got this lovely mail today (sender shall remain nameless and not possible to identify as I have no intention to point fingers). The mail doesn’t mention any project I’m involved in, nor does it say how the guy gets his problem or anything. The question is simply:

I hope that you managed to resolve Segmentation fault. I am having the segmentation problem when I run my Executable, I think it got to do with Remote failure Reply, I am not sure. My executable is build under ARM-Linux. I will be grateful if you can tell me which Library file you included or deleted in order to get rid Of segmentation fault.

I love it. I have no idea what he’s talking about..! Quite likely everything will be evident once just a tad bit more details are revealed.

Linux Sound, part II

Here’s a follow-up on my previous rant about Linux sound still lacking:

I got the advice to try shutting off my on-board audio in the BIOS settings, and indeed I think that’s a great idea as with only one sound card present in the system I figure the chances are much bigger that things would auto-detect and work smoothly.Music Note

Forget that. When I did that, I failed to get any sound at all. Not by default and not after trying one of my alsaconf and aumix tricks. I went back and enabled the on-board sound again, did the insane command line invokes and voila, sound is now back…

Someone mentioned that this situation of course may be different with different distros, and while that may of course be entirely true it really doesn’t matter much to me since I won’t change distro (now) and I would expect Debian to be at least around the distros that do a half-decent job.

tech, open source and networking