Forgot your password?
typodupeerror

Comment: Re:Not news (Score 1) 239

by dalias (#48125167) Attached to: Where Intel Processors Fail At Math (Again)
If you bothered to read the story, the problem is that the result is not correct to 1 ULP. Even for small inputs it has significant errors and the worst case has only 4 bits correct (so, roughly 2^49 ULP error). As for deprecation, you're right and wrong. It's not officially deprecated, but it's slower than computing sin correctly in software, and between the performance issue and the fact that experts in the field have known about this bug for a long time, nobody with a clue uses the FSIN instruction.

Comment: There is a real problem (Score 5, Informative) 345

by dalias (#48124119) Attached to: ChromeOS Will No Longer Support Ext2/3/4 On External Drives/SD Cards
Ext 2/3/4 and any filesystem that records file ownership (especially numeric uids/gids) is not suitable for storage that's not associated with a particular system's user account database (/etc/passwd or otherwise). Linux could attempt to support such usage by virtualizing/remapping uids for "external" ext2/3/4-formatted drives, but it doesn't. Instead, you have a situation where file ownership is just silently wrong when you plug the drive into a different computer. So removing support is a big hammer, but I see how they could see it as a justifiable one when the status quo is broken like this.

Comment: Why?? (Score 1) 109

by dalias (#47110181) Attached to: Firefox OS Powered Flame Available For Pre-order; Ships Globally
Why is anyone still making devices with sub-300-dpi screens? Especially when you're trying to launch a new OS -- this is the best way to make your OS look like crap, even if it's otherwise great. Price is definitely not a show-stopper here; the ASUS ZenFone has a high-end screen in the ~$200 range, and cheap Chinese phones like Zopo have had them for a long time at much lower prices. If they're really worried about cost, scrap the 5 MP camera which is a complete joke when your screen is 0.4 MP...

Comment: Re:Overpriced at $0.60 (Score 3, Insightful) 89

by dalias (#46760559) Attached to: Paper Microscope Magnifies Objects 2100 Times and Costs Less Than $1
Toy microscopes don't work at all. Their focus knobs are loose so that you constantly lose focus while trying to see the sample, and they only have one focus knob which makes it essentially impossible to focus to begin with (real microscopes have coarse and fine knobs). And the magnification rating is always fake. If they advertise 400x, expect resolving power so poor that they're essentially 20x or less. I once got one of these pieces of junk and ended up going back to eBay for a $80 vintage Bausch and Lomb scope which I'm very happy with, but sadly I think that was a rare find and I just got lucky.

Comment: Re:Famous last words (Score 2, Insightful) 179

by dalias (#46677269) Attached to: "Nearly Unbreakable" Encryption Scheme Inspired By Human Biology
"Climate change" is not a "downgrade" to global warming. It's simply better wording to avoid denial from idiots who don't understand math (i.e. means) and say "wow it's really cold this winter, global warming is bs!" Nothing has changed; we still know the mean temperature is increasing and that the increase is caused by human activity. But the new wording is less susceptible to idiotic misinterpretation.

Comment: Re:Opensource and web services keys (Score 1) 109

by dalias (#46564601) Attached to: AWS Urges Devs To Scrub Secret Keys From GitHub

If your FOSS application interacts with a web-based service that requires an API key, the correct way to implement it is to instead have it interact with your own servers, and in turn have your servers interact with the web service via the API key. You should of course then publish the source to the server-side part of your application as well, and advanced users can then (if they really want to) setup their own server, with their own API key for the web service; this also protects users from the possibility that you disappear and shtudown your server or let it rot.

Of course this design assumes it's a web service your users are accessing anonymously. If they have to login to their own accounts, then this model is usually wrong. They should never be providing their account credentials to you, and it can only work correctly with more advanced authentication methods that avoid the need for them to provide credentials to you, which the web service is unlikely to support.

Comment: Re:Compared to Bionic (Score 1) 134

by dalias (#46539733) Attached to: GNU C Library Alternative Musl Libc Hits 1.0 Milestone

I really want to add a Bionic comparison, but in order to be comparing apples with apples (or non-apples with non-apples, pardon the pun) we need an x86 build of Bionic, or need to re-do all the other libcs' figures for arm. I've been looking for a way to build Bionic outside of the Android build system and use it on non-Android systems, and the gentoobionic repository at https://github.com/gentoobioni... looked promising, but I couldn't get it to work. It also may be much larger than the official Bionic.

If anyone is willing to help us figure out how to setup x86 Bionic for testing, please stop by the IRC channel (#musl on Freenode) or send a message to the mailing list.

Comment: Re:Either gnu libc is hideously slow and bloated.. (Score 1) 134

by dalias (#46539631) Attached to: GNU C Library Alternative Musl Libc Hits 1.0 Milestone
Someone with mod points, please mod up the parent post. Even if you disagree with it, it's informative about one of the big issues in glibc that musl does differently: musl's snprintf and dprintf, for example, are async-signal-safe. Roland McGrath, who holds claim to being the "inventor" of dprintf and author of the original implementation in glibc, has stated that he intended for the function to be async-signal-safe or at least close to it, and that later introduction of dynamic allocation is a bug (which I later filed as #16060) that glibc should fix.

Comment: Re:pkgsrc test results (Score 1) 134

by dalias (#46539393) Attached to: GNU C Library Alternative Musl Libc Hits 1.0 Milestone
There is no isinf symbol or reference to one in musl, so I think this is something to do with your toolchain (a BSD-packaged version of LLVM that tries to make itself look like gcc?). Pretty much all of your concerns (including "lack of C++") could be solved by building a toolchain to target musl (note: uClibc and glibc make you do this anyway; musl is fairly unique in providing a way, albeit sometimes clunky, to use the new libc without a new compiler toolchain). If you want to do that or proceed trying to get the wrapper to work on your system, you'll find people who can help in Freenode #musl or on the mailing list. On the other hand I understand if you don't want to go to the trouble, but keep in mind you're using a non-native setup that's probably never been tested.

Comment: Re:pkgsrc test results (Score 1) 134

by dalias (#46536881) Attached to: GNU C Library Alternative Musl Libc Hits 1.0 Milestone
The problem appears to be that "x86_64" is identified by your compiler as "amd64" in the machine tuple. This is easily fixed by adding --target=x86_64 (the name musl knows it by) on the configure command line. I don't see any reason we can't add "amd64" to the detection logic in configure too, so this should work better for you in a future release.

Programmers do it bit by bit.

Working...