Diary for Jan. 2002


daily blurbs

30 Jan 2002

00:05:19 people using jed and mutt are experiencing problems with slang1-utf8. It seems to be causing a lot of grief, and other things. Too painful. Too bad.

00:31:45 I found out what libslang was trying to do. It is a pity that it failed, but really, it is annoying.

02:22:05 Hacking slang is taking too long. I think I got things somewhat right. Sonames are now specified properly, I think. I am not quite sure.

22:14:20 I have noticed that slang fix will break the newt build process. I am going to take a look at it, and see if I can fix it. Wow. I know of one way to fix it, make libslang-utf8.so symlinked to libslang.so.1-UTF8.4.4, and be done with it. Or maybe use that thing in newt build process, but it is too hackish, and it will break in the future. Maybe a slang-both-dev for building for both ? I don't know. Avoiding building newt-utf8 in the newt build process would solve many problems, I think. Then it can go on to building newt-utf8 source package which can build newt-utf8 libraries. I am not sure.

23:37:23 newt will need to link with slang with some kind of -Bsymbollic flag. Because slang is incompatible. Or just stop lying and make two versions of newt, one with utf8 support, and the other without.

29 Jan 2002

18:17:52 I've discovered that slang1 and slang1-utf8 are not compatible. Bad idea, well, at least I know now.

28 Jan 2002

02:58:26 adopted mtools package. Browsed through the upstream patch, and it seems only necessary for cygwin. I think I should ignore that, unless cygwin port of Debian becomes so necessary.

26 Jan 2002

01:45:10 I'm running on my crusoe machine, and I can get the load average to around 0.3 by just running some interactive programs. I am just checking my mails, chatting on IRC, and using emacs. That doesn't make visible load on an Athlon, but it does seem to make a difference to Crusoe machine.

22:31:05 I've wrote a very short patch to make sshd have configurable version string. I don't know if it really works, because I can't test it. Trying to build this code requires too many... However, this patch probably won't be applied.

Patch from Junichi Uekawa <dancer@debian.org> to allow for
random version strings depending on SSHD_VERSION env var.

diff -ru openssh-3.0.2p1/sshd.c openssh-3.0.2p1-patched/sshd.c
--- openssh-3.0.2p1/sshd.c	Sat Jan 26 22:08:44 2002
+++ openssh-3.0.2p1-patched/sshd.c	Sat Jan 26 22:07:55 2002
@@ -330,7 +330,10 @@
 		major = PROTOCOL_MAJOR_1;
 		minor = PROTOCOL_MINOR_1;
-	snprintf(buf, sizeof buf, "SSH-%d.%d-%.100s\n", major, minor, SSH_VERSION);
+	if (getenv("SSHD_VERSION"))
+	  snprintf(buf, sizeof buf, "SSH-%d.%d-%.100s\n", major, minor, getenv("SSHD_VERSION"));
+	else
+	  snprintf(buf, sizeof buf, "SSH-%d.%d-%.100s\n", major, minor, SSH_VERSION);
 	server_version_string = xstrdup(buf);
 	if (client_version_string == NULL) {

25 Jan 2002

03:12:31 I've ran from my lab to home. It took me one hour. Not too bad.

17:04:49 I took a glance at ltrace. The source had some non-pretty parts in it, which I wanted to get fixed. I filed a bug report with a patch. I have some more things in my mind that I want to get fixed, but this things has a lower priority.

21 Jan 2002

16:10:35 My personal load average is very high, and I think things are not going pretty well, because things are not finishing by the time they should be. I am in a very very BIG TROUBLE!!

20 Jan 2002

21:08:02 Checking out openoffice. The last Debian package, 638, has a few problems: it needs libgpcl0 which is non-free, and also requires JAVA, which is non-free. Bad news. The progress was in the direction of removing java requirement in the upstream, but the guy working on it has been suffering some lack of HDD space, or so I heard.

19 Jan 2002

11:03:23 My deadlines are approaching very rapidly. However, I have uploaded pbuilder despite the fact that I severely lack time.

18 Jan 2002

17:29:30 Pretty funky day. I haven't slept much for sure... Tried to apply the pbuilder patch. I think it fixes many problems that was experienced previously. Funny. It's important to re-read the code. It's very nice to have some peer-review of code by someone else who is quite knowledgeable. I've finally managed to upload pbuilder. It should be installed sometime in about 12 hours. Test it if you wish.

17 Jan 2002

11:07:52 I celebrated the Annual for Debian Developer last night, with the Debian users around me. There was a question about "how many debian users are there?" and I answered that it is impossible to know. How many, really ? Do we have a way of knowing ?

15 Jan 2002

18:35:26 Somebody was shouting about diskless, and I noticed that software page listed it as one of the "unmaintained" packages. Well, the adoption did not take place, yet, and I am still maintaining the package. If you want it, contact me. Before playing.

14 Jan 2002

02:30:45 Patch for wajig. The maintainer for wajig was rather annoyed by my bug filing. It is interesting. Python is a very mysterious language to me, and playing with it does not give me much pleasure. Not much so far, yet, anyway. I'm probably too old-styled to see the benefits.

diff -ru wajig-0.2.11-orig/src/Makefile.in wajig-0.2.11/src/Makefile.in
--- wajig-0.2.11-orig/src/Makefile.in	Sat Oct 27 19:25:06 2001
+++ wajig-0.2.11/src/Makefile.in	Mon Jan 14 02:21:30 2002
@@ -10,6 +10,7 @@
 srcdir		= @srcdir@
 pycomp		= @top_srcdir@/py-compile
 CFLAGS          = -fPIC -shared -O @PYTHON_INCLUDES@
+wajiglibdir     = @libdir@/@PACKAGE@
@@ -42,8 +43,8 @@
 	${pycomp} *.py
-	${INSTALL} -d ${datadir}
-	${INSTALL} -m 644 *.py *.pyo ${datadir}
+	${INSTALL} -d ${wajiglibdir}
+	${INSTALL} -m 644 *.py *.pyo ${wajiglibdir}
 	rm -f ${datadir}/*.py ${datadir}/*.pyo
diff -ru wajig-0.2.11-orig/wajig.sh.in wajig-0.2.11/wajig.sh.in
--- wajig-0.2.11-orig/wajig.sh.in	Sat Oct 27 19:25:06 2001
+++ wajig-0.2.11/wajig.sh.in	Mon Jan 14 02:23:58 2002
@@ -3,7 +3,7 @@
-export WAJIGDIR=@datadir@/@PACKAGE@
+export WAJIGDIR=@libdir@/@PACKAGE@
 @PYTHON@ $WAJIGDIR/wajig.py $*
Only in wajig-0.2.11: wajig.sh.in.modified


12 Jan 2002

13:59:47 I'm needing some free time. I'm going to write two papers. That's hard. I don't know.

21:01:04 I had a pretty good 30 minutes or so with my girlfriend. I like the mood, when she is so cute and cuddly. Yes. It is a joy.

10 Jan 2002

13:30:32 I've started to write a little bit on shared library guide. It should be going on quite well. I am not sure. Because I am seeing that there are too many people without much clue, and do not understand the manual. There are many good signs, such as "do not match the soname with the version number of the package", and I think many people are missing it. I am not quite sure of some points, and discussing some core points about some reasonings of obscure points will most definitely help me understand more about all.

22:17:05 I'm rather depressed right now. I'm feeling anger. This is not a good sign, I drank too much last night, maybe, and I have not recovered. This is no good news.

23:33:45 upload ecasound. A very small fix. I think doing too big a change does not help at all. I was writing up a little text about libraries, and other things. Maybe I don't know enough to write one. Comments welcome. Well, I probably need to post it somewhere public, but I want to ponder over it a little bit more.

9 Jan 2002

02:17:09 Unsubscribed from alsa-devel and alsa-user. I don't really feel related to the discussion there. I'm impressed that ALSA is being ran by so few people. It's pretty interesting.

8 Jan 2002

00:46:47 pbuilder 0.17 is uploaded to Debian now. It should be quite fine.

15:27:52 I realized the submission deadline for the paper is coming very very close by.

7 Jan 2002

00:38:19 TODO: Add to danceracpi some gdk_window_set_hints. Add rpm support to apt-el.

15:38:05 fixed dmachinemon slightly, to fix dmscripts bug.

6 Jan 2002

00:17:15 I have updated the software page for apt-mode. It should be quite functional.

5 Jan 2002

01:05:10 Fixed danceracpi so that it will check for existence of /proc/acpi. This is good.

14:28:11 I have written a small script to parse debian/control file and output some XML file. I think it is pretty neat. I like it anyway. Better than other things that seem to be pretty proprietary and difficult.

14:41:49 apt-el has some features to install packages now. This is good, getting closer to the functionalities of dselect. I now need to get some highlighting and other minor things to be done. I probably will be better off with some help from some guru.

17:05:47 Adding version information to danceracpi, so that it can be released to the public without problems. It should be okay.

21:01:46 python-xml problem for boot-floppies is fixed. This is good. Collaboration is working fine. MishaS has probably fixed it...

4 Jan 2002

00:36:23 I am considering the following: The policy regarding Debian library packaging needs to be fixed. The development library must be named lib[PACKAGE][SONAME]-dev which Provides: and Conflicts: lib[PACKAGE]-dev. The packages which use the development library must Build-Depend on lib[PACKAGE][SONAME]-dev. Anything else is load of crap. Maybe I am dreaming. Give me a good reason why lib[PACKAGE][SONAME]-dev idea is not good. We have a choice currently, but the choice is not done quite correctly. It is not obvious, but specifying lib[PACKAGE]-dev in the build-depends does not ensure that all architectures have the same library compiled against a binary. This thing is getting more and more of a problem due to "testing" distribution being used.

18:53:59 Changed the key binding for dancer-diary.el to make it to C-cC-j. It seems to be more compliant with other emacsen.

20:51:13 Some guy has sent me a question regarding dsh and NIS support. I should say, that sending questions to the upstream author without checking anything, sending "suggestions" is useless. It might not even be the case. If you don't use the code, it is really impossible to know what is lacking. This is probably the weakness of open source systems in general, there are alternatives, and not much motivation to integrate the different alternatives. I need to check if netgroups can be managed in an hiergraphical manner. If that is possible it would be feasible to create information so that it will reflect the real structure of cluster when it has a really hierarchical system. By the way, I don't use NIS to set up my cluster. I am looking at other ways.

3 Jan 2002

00:20:13 I fixed a memory leak in libdanceracpi and it seems to be stable now. It's pretty interesting that a usable system can be constructed in a very short time. Text-based system is a very pretty thing to have, I think.

10:47:02 There is a boot-floppies bug with python2.2-xml. This is pretty bad, because it is about python, and there are not many people who can grok python. Hmm... I kind of know how to fix this. Thanks Je'rome.

23:33:27 Updated the software pages. I guess it is important to record what I am "trying" to do, so that I don't lose myself.

2 Jan 2002

13:09:10 Chatting with manoj about the boot-floppies, and its bugs. It is not running okay. /usr/bin/debootstrap --arch i386 --print-debs woody gives me the list of packages that are considered base. They need to be on the first CD. The following script kind of checks what is wrong.

	for A in $(/usr/sbin/debootstrap --arch ppc --print-debs woody); do (apt-cache show $A | awk '/^Package:/{package=$2} /^Section:/{section=$2} /^Priority:/{priority=$2} END{print package " " section " " priority}') ; done | awk '{if ($2 != "base" && $3 != "required") {print "Bad base program: " $0 }} '

13:09:33 X seemed to have magically crashed on this system. Bad news.

16:01:15 My room is too cold to do anything decent in it. Freezing. January is a nice time of year.

1 Jan 2002

17:17:40 Things are improving, this is a new year. Libretto is getting good. I am furiously coding a lot of things. I need to write up some things, but I am not quite in the mood yet. Well.. sometime.

18:31:21 (insert-image (create-image "full-path")) seems to insert an image into emacs21 frame. This is cool.

23:05:51 Started to write an ACPI battery monitoring system software. There are not many of them around, and I wanted to create something feasible and usable. I have got some kind of libdanceracpi and some text mode application working. I should get things going. With the past experiences with dmachinemon I should be doing better now. But nobody knows. Using a shared library from the start is a good start. Rearranging the contents of the shared library is easy, as long as other applications do not depend on that application. Code sharing is important. Copy and paste does not last too long. I need a better way than to copy and paste, which already exists: the shared library.

Junichi Uekawa

$Id: 200201.html.en,v 1.41 2002/02/01 18:20:40 dancer Exp $