On Wed, Nov 28, 2007 at 11:26:39AM -0500, John E. Davis wrote: > On Wed, 28 Nov 2007 16:19:48 +0100, Remko van der Vossen <wich@xxxxxxxx> said: > >> slang version is 2.1.3 > >> xterminal is mlterm version 2.9.3 > >> LANG is set to ja_JP.UTF-8 > >> glibc version 2.4 > > > >It seems to be a problem in combination with glibc, when I try the same > >binary on a different system with the same versions of slang and mlterm > >and the same LANG but glibc version 2.6.1 the problem does not occur. > > > >Now the question is if this is a bug purely in glibc or does slang do > >something that doesn't agree with glibc 2.4. Does slang support glibc > >2.4? > > I have not heard of any slang-related issues with glibc 2.4. Did you > compile slang for glibc 2.4? Does the problem occur if you use xterm? > Does a C version of you code show the same problems? How do the > smgtest and pager programs distributed in the slang/demo directory > perform? My observation about glibc seems incorrect, the testing environment was not the same. I ran the original test in screen and the test on the glibc 2.6.1 machine outside of a screen. Further testing seems to show glibc 2.4 or 2.6.1 does not matter (much). I have not yet tried a C versions of the code. An overview: case glibc xterm screen app 01 2.4 mlterm yes test fails 02 2.4 mlterm no test works 03 2.4 xterm yes test works 04 2.4 xterm no test works 05 2.4 uxterm yes test works 06 2.6.1 mlterm yes test fails, but better than case 01 07 2.6.1 mlterm no test works 08 2.4 mlterm yes smgtest act fails *1 09 2.4 mlterm no smgtest act fails *1 10 2.4 uxterm yes smgtest works 11 2.4 mlterm yes pager works 12 2.4 mlterm no pager works screen is version 4.0.3 on both machines *1 The "Another Color Test" fails, the first screen (black, white, green, red in clockwise order) is incorrect (no white) the second, third and forth screens in the cycle appear to be correct. (All other tests, notably the wrapped string and monochrome tests, seem to work fine.) btw, after the BCE color test the background does not switch back to black which might impact some of the other tests, it might be wise to reset it to blak after the test. For these tests it did not seem to matter, I tried all tests with both a black and blue background with the same results. Judging from the tests with just my test program it would seem to be an issue with mlterm and screen, but I have no such problems with any n(w)curses based programs that are on the same system... I don't know if I have any slang programs that could exhibit the same problems. Let me know if you want me to test some specific other combinations, I will strip the c++ from my test program and see if that yields the same results and let you know about it. Hopefully you can reproduce the problem with the above information. Regards, Remko van der Vossen
Attachment:
smime.p7s
Description: S/MIME cryptographic signature