Comments on: Debian is switching to EGLIBC

By: glibc timezone integer overflow | Web Security Watch

glibc timezone integer overflow | Web Security Watch — Tue, 07 May 2013 17:52:54 +0000

[...] is even more friendly, cooperative, and socially well-adapted than a certain OpenBSD maintainer. http://blog.aurel32.net/?p=47 illustrates this [...]

By: Debian 也向 Embedded 看齐！换上 EGLIBC！ | IT档案馆

Debian 也向 Embedded 看齐！换上 EGLIBC！ | IT档案馆 — Tue, 29 Jan 2013 09:35:29 +0000

[...] Jarno“在 blog 上发表了一文＜Debian is switching to EGLIBC＞，文中宣告了 Debian 将会逐渐使用＜EGLIBC(Embedded GLIBC)＞取代 GLIBC(GNU C [...]

By: Open Source: The Scientific Community in Technology | Stephan Sokolow's Blog

Tue, 05 Jun 2012 11:17:56 +0000

[...] Jarno, Aurelien. Debian is switching to EGLIBC. (May 5, 2009). Aurelien’s weblog. Retrieved October 9, [...]

By: S02E04 – Hidden Danger – OGG HIGH | Ubuntu Podcast

S02E04 – Hidden Danger – OGG HIGH | Ubuntu Podcast — Mon, 27 Feb 2012 18:16:34 +0000

[...] Battle of the GLIBCs! [...]

By: Sysadmin Haiku » Debian Squeeze congelada

Sysadmin Haiku » Debian Squeeze congelada — Fri, 25 Mar 2011 22:27:14 +0000

[...] que ya tendría que haber finalizado la migración de glibc a eglibc, otra de las novedades que nos contaba Auréllian Jarno en su blog, y la introducción de Startup, la variante nueva del sistema de inicio basada en el [...]

By: Isolated from the Outside World « This week in panospace

Isolated from the Outside World « This week in panospace — Sat, 26 Feb 2011 03:24:41 +0000

[...] switch from GLIBC to EGLIBC affected functionality of some software like Skype for a few hours, but I [...]

By: Aurelien’s weblog » Blog Archive » EGLIBC and PowerPCSPE port

Aurelien’s weblog » Blog Archive » EGLIBC and PowerPCSPE port — Tue, 25 May 2010 15:57:32 +0000

[...] has been roughly one year since Debian has switched from GLIBC to EGLIBC, so it’s probably the time to make a small report about this [...]

By: Tony

Tony — Fri, 26 Jun 2009 19:38:02 +0000

I’ve also updated the NPTL Wikipedia entry, in order to give information about the PTT tool and about he OPTS (Open POSIX Test Suite) tool for testing NPTL against the POSIX standards.
Sébasitien Decugis, who worked under my direction at Bull SAS, spent a lot of time for implementing OPTS, which provides conformance, functional, and stress testing, with main focus on Threads, Clocks & Timers, Signals, Message Queues, and Semaphores.
OPTS can help you to check that eglibc works fine. We found several problems in NPTL with it.
Tony

By: Tony

Tony — Fri, 26 Jun 2009 19:21:30 +0000

I’ve updated the French version of Drepper Wikipedia, so that the information about the transition of Debian from glibc to eglibc does appear.
I’ve also added a note in English and French version of Wikipedia about the fact that Ulrich Drepper refused to add any tracing tool for the NPTL thread library, though several people asked me during the last 3 years why PTT is not part of the glibc.

BTW, if you think that such a tracing tool would be useful for eglibc, go and take it. Everything were done to make the adding of PTT to the glibc as easy as possible.
Working now in the AIX world, I see how important it is for our customers to be able to trace AIX components.
Bests,
Tony

By: Tony Reix

Tony Reix — Fri, 26 Jun 2009 11:56:25 +0000

Hi,

I’ve just discovered this switch from glibc to eglibc (I no more work in Linux world. Now back to AIX since 3 years).

I fought some years ago with this ugly ass-hole egocentric Ulrich Drepper when I proposed him to incorporate into the glibc our PTT (Posix Thread Tracing Tool) tool designed for tracing the NPTL routines. Around 24 p.m of work. Previously I develop a POSIX thread test suite that found some problems in NPTL.

See: http://nptltracetool.sourceforge.net/

Read what I said this morning to a guy asking if PTT is still alive:

The project PTT has reached its goals:
1) have a high quality and very good performance,
2) has nearly no performance impact when not used, and
2) be ready to be integrated by the libc guys.

However, the maintainers of the glibc (including Ulrich Drepper) were non-professional guys at that time: they refused to integrate our work (me + 5 students, about 24 p.m of work) in the glibc.

The reason ?
1) These guys thought that their wonderful code had no bug….
2) They never worked in the real world and they never understood how important it is for the end-users/customers to be able to capture
information about why things does not run as we expect they should. And any non-stupid guy knows that programs never work as we expect they should.
3) They refused to integrate code they didn’t write (understand that no one can write so marvelous code as they do…).
4) They didn’t want to have some more code to maintain (understand that they would had to spend some time to read code they didn’t write).

Now that Ulrich Drepper works for RedHat, I guess he has learned a
little bit more about the debug/tracing needs of customers: RAS (Reliability, Availability, Serviceability). However, I think that such an ass-hole guy can never change his mind since he is so egocentric.

I see that this (bad) guy has a Wikipedia entry. I’ll add some nice
comments one of these days.

If Linux kernel does not plan to include more trace/debug tools, Linux will never be a professional tool. I mean that, for big critical
projects, customers require to have tools to capture failures at the
time they appear or to be able to capture data for understanding what happens.
I now work again in the AIX world and I see how important it is for our customers to be able to capture failures.

So you are free to get the source and port it to the last version of
glibc.
I think that my students provided enough documentation to help people to port it easily, though you must spend time to understand how NPTL works. If you plan to do that, you may ask Guillaume if he can provide you with advice.

Regards,

Tony
Senior AIX System Management Architect