Description of problem: Glibc 2.3.3 is the first glibc with IDN support (umlaut domains) using GNU libidn, but Red Hat Enterprise Linux 3 has glibc 2.3.2 - also it hasn't IDN support. IDN is a upcoming thing, especially in Europe. Many production systems and systems used for application development are running Red Hat Enterprise Linux 3. As long as RHEL4 is far away (~ 1.5 years), there should be a backport for RHEL3 - maybe for U4?! I think, the system administrators and developers using RHEL3 should get the chance to have, experiment and develop things using IDN already, because Fedora Core 2/3 (which have currently working IDN support) aren't normally used in critical enviroments. As I talked with Jakub Jelinek, he said that it is pretty obvious and libidn is a separate add-on and the changes to core glibc files are really minimal. Version-Release number of selected component (if applicable): glibc-2.3.2-95.20 Actual results: I backported the IDN support from glibc 2.3.3 back to 2.3.2, but I had to made some changes outside the IDN specific things, because there was development and some code changed. I talked with Jakub Jelinek (I think you're the default but owner, too) - and sent him the "critical" part of my changes in "inet/getnameinfo.c" for review. Expected results: glibc for RHEL3 with IDN support... ;-) Additional info: I would append my IDN patch worked out against glibc-2.3.2-95.20, if Jakub or someone else approves my code change in inet/getnameinfo.c, where I'm really not sure whether it is wrong or right what I did.
Created attachment 102426 [details] glibc-2.3.2-200309260658-idn.patch This attached patch for IDN support in glibc from Red Hat Enterprise Linux 3 works for me fine since 2004-07-15 at all my systems (I still love the "risk"), even if no Red Hat or glibc guy reviewed it, yet...
Internal RFE bug #129479 entered; will be considered for future releases.
This is not feasible for a RHEL 3 update release.
Is it possible to get a good reason for that totally absurd decision? The IDN "[libidn] support is a separate add-on and the changes to core glibc files are really minimal" Jakub approved, so I guessed it shouldn't be a big problem. Suzanne, let me say it more clearly (I still think a typical American can't see reasons for IDN support, I already noticed that at other parts - especially at non-european companies): Red Hat, Inc. also has customers in Europe where IDN (umlaut) domains are currently coming more and more and they are even more of belong as you can think. Last but not least: Sorry for the hard and brutal words against you and Red Hat, Inc. but I want to be direct and open - maybe you can understand that - if not the rest?!
The point was primarily that a glibc change doesn't make much difference, every application, that does a host name lookup and needs to handle IDN domains, must be changed. And that is not going to happen for RHEL3.
Jakub, I never thought and talked about using IDN support in current and existing applications of RHEL3 (maybe you read again, what I wrote?)...but it's okay - I know what I've to think and I'll use this patch further one here without Red Hat.