A frog around the world

@teh 1

> what's the general reason why everything won't compile on Clang?
There are various reasons:
* bad code that gcc accepts
* non standard C code accepted by gcc (variable length array for example)
* -Wall (+ -Werror) does not enable the same set of warnings
* clang is C99 by default, not gcc.
* etc

Out of pure intellectual curiosity, what's the general reason why everything won't compile on Clang? Is it that GCC ignores errors and alows non-standard code? Or is it Clang that didn't implement everything according to standards? (In terms of C and C++ standards)

You might be interested in the Gentoo tracker for Clang-specific failures:

https://bugs.gentoo.org/show_bug.cgi?id=408963

Amazing decrease from 9.5% to 5.7%.
I guess it will be possible to compile a full Linux distribution with clang in only a few years.

Very nice work Sylvestre.
:)

Aaron, Planet Debian has already an account:
https://twitter.com/planetdebian

Vincent, why not but it would help to get that in a better format (RSS, XML or RFC 822 format).

A Debian Planet feed would be great also. I don't know if this exists already, however.

Having an account for new Debian maintainer [1] would be cool, I guess!

[1] https://udd.debian.org/cgi-bin/new-maintainers.cgi

Thank you for doing this. I see that a lot of people find it confusing why there isnt a change log between beta releases.

http://gcc.gnu.org/ml/gcc/2014-01/msg00247.html

Amazon gives 10 000 US$/year to Debian for such experiment.

We are building, for now, just for amd64. There are still so much work to do. It would be a mistake to try tackling every issues at the same time.

What is "AWS donation" ?

Are you only building x86 or other arch too ?

Sorry, I missed your comment.

I am not planning to do that soon but that would be nice.
By the way, clang is already the compiler used to be Firefox under Mac OS X.

Thx again,
I wish I could help but I have no time right now.

clang-modernize-3.5_3.5~svn202666-1~exp1_amd64.deb 03-Mar-2014 09:37 3.1M
clang-modernize-3.5_3.5~svn202688-1~exp1_i386.deb 03-Mar-2014 09:37 3.3M

Why this two have different version numbers?

@Seb: For now, you cannot install both clang 3.4 & 3.5.
I reported this bug:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=736057
and I will be happy to mentor someone to fix that bug.

Thank you for this update.
Somehow I can't figure out how to install clang 3.4 and 3.5 in parallel.
When I try installing clang 3.5 apt says it will replace clang-3.4 (same for clang-modernize-3.4 lldb-3.4)
I run on ubuntu saucy.
Do you know if can be done ?

Hi, Sylvestre.

Congrats on the new job. Just a speculative thing: do you plan on experimenting with compilation/speed comparisons of firefox/iceweasel with clang against gcc?

I guess that this would be a super nice thing to know about the browsers.

Before I do something like this, would you anticipate any improvement at all from excluding features from firefox (e.g., accessibility features, parental controls etc.) as is advertised by this unofficial build for windows:

http://www.palemoon.org/technical.shtml

?

Thanks for any insight,
Rogério.

any plans for a test rebuild with clang 3.4?

Thanks much for the mention. Useful tool.

Yes: --enable-experimental-targets=R600

Search for ../configure in:
http://anonscm.debian.org/viewvc/pkg-llvm/llvm-toolchain/branches/snapshot/debian/rules?view=markup for the list of argument

Does Radeon backend is enabled?

"even if the error is obvious", make that plural. You can see it in the error report: "..is located 5 bytes inside of 80-byte region..". Why an 80 byte region for a 10 char array? Because it's allocating 10 char pointers, which, on 64-bit architecture, are 8 byte. The malloc should get (10 * sizeof(char)).

Just as long as we don't get loads of half baked patches to "quieten AddressSanitizer" now (a la OpenSSL).

It is a Google development, and it will also be available in GCC 4.8

@MJ Ray: Fairtrade, I do not know, which is not particularly reassuring, but the short-sleeved polos are indicated to respect at least a textile quality standard called ?Eco-Tex?. The long-sleeved ones are not indicated to respect any specific standard though?

MJ Ray: I don't know, sorry :/

Is the cloth fairtrade? Does it avoid sweatshop cotton?

I wrote about this a while ago in http://www.news.software.coop/are-your-free-software-t-shirts-ethical/112/ and I'd be interested in a bluff if it's ethical and available to ship to England.

Oh and how do I tell this site to email me on new comments?

For those of you that may notice the very low quantity of rugby shirts we will have for sale, this is by design: as this kind of cloth is not very usual in our kind of community, and quite more expensive than the usual tee-shirts, we chose to get only a few of them to test how they will be appreciated.

Now, if it appears they are appreciated, do not worry, we will have more next time! :-)

@Julien, it works for me ?! (both reader + planet)

Nice post, but something is wrong with the planet.debian.org integration or with Google reader, as that leads me to

http://sylvestre.ledru.info/blog/xmlsrv/805@http://sylvestre.ledru.info/blog

(your guid) when trying to open the blog, as reader seems to ignore and only respect elements.

Congratulations! This is very helpful to those of us developing with C++11 and Clang on Debian!

Valio ??

Rss encoding is broken :(

I think that the number of commits is multiplied by 3.
Also, "Eli Friedman" is with Apple. "Owen Anderson" is with Apple. "Bruno Cardoso Lopes" is academic, AFAIK. "Reid Spencer" did most of his work before joining AutoESL (which is now Xilinx, I think).

The desperation in your voice is funny point.

I think you know perfectly well what the point is and your response proves it is valid!

What is your point? LIcense is owned by U of Illinois Code owners are all employed by Apple. None of the Google employees have contributed any significant code aside from Richard Smith. Apple also owns WebKit tradmark, OpenCL, libdispath, Objective C. Once again what is your point?

No, I didn't. I am using the email addresses registered in the LLVM VCS.

Did you also account for change of employment for those @gmail, etc.?

Excellent. I have been trying out clang for a while as a replacement for gcc. It would be great to see better support for it in Debian.

Can you provide search by package name or source package name on clang.debian.net? It will make easier to check if packages I care is affected. Also package size comparission (before and after compressing into .deb, and default ones from Debian) of this packages which successed would be usefull.

Another question: what architecture you was using? amd64? How about i386? It should be equally easy (in terms of access to building virtual machines), and when considering metrics like package (binary) sizes changes beetween clang and gcc, it can give better insight into any changes.

the sun-java6 package seems also to be updated/available here :

https://github.com/rraptorr/sun-java6

This is… well, not great, but useful. Not great because, in my heart, I’m fully behind throwing it out. Useful because, to update mksh in Android AOSP, I need to be able to build and test-boot it – and Google is still depending on Sun J2SDK 6 for that.

I tried lenny’s package, but first had to patch it to accept two-digit patchlevels and amd64 packages, and THEN ran into the problem that update-java-alternatives didn’t like the result. At all. If this tool version does it better, I’m bought.

Merci!

Julian: apt-get install icedtea-web
See:
http://packages.qa.debian.org/i/icedtea-web.html

Not only corporations, also everyone wanting to use applets basically needs Oracle's Java build, as the free plugin basically does not work with many sites.

Thx for bringing java-package back!
Btw... i update the sun-java6 package regularly to make it work with the newest java 6 version.

https://github.com/gebi/sun-java6

Ubuntu is regularly synchronizing against Debian... So, it should come sooner or later.

any plans to include this into ubuntu?

All this whining about LLVM/Clang. What a joke. The quality of architecture alone, nevermind the contributions by hundreds of architect level quality developers of compiler expertise is producing a far superior product to GCC.

You get all of this for free.

You get CUPS for free. Nothing is stopping any Distribution of Linux to pair up the quality of CUPS with that of Apple.

dd-list will be cool addition :)