MedWorm.com provides a medical RSS filtering service. Over 5000 RSS medical sources are combined and output via different filters. This feed contains the latest items from the 'Eugene Melamud's Blog' source. Sat, 16 Aug 2008 14:41:52 +0100 http://emelamud.blogspot.com/2007/03/simple-raid5-recovery-on-linux.html Step 1. Detect failed disk [root@hassam backup]# more /proc/mdstat Personalities : [raid5] read_ahead 1024 sectorsEvent: 1 md0 : active raid5 hdb1[0] hdc1[1] hdd1[2] hdi1[3] hdj1[4] hdl1[6] 1465175424 blocks level 5, 64k chunk, algorithm 0 [7/6] [UUUUU_U]see the missing U there, that means missing or non functional disk.Step2. Identify failed disk. Lets find out whats wrong. First we will look into /var/log/dmesgs and /var/log/messages to find out if there are any clues there. (grep for md, to simply this). You might see something like this. md: running: />md: hdl1's event counter: 0000002dmd: hdk1's event counter: 00000028md: hdj1's event counter: 0000002dmd: hdi1's event counter: 0000002dmd: hdd1's event counter: 0000002dmd: hdc1's event counter: 0000002dmd: hdb1's event counter: 0000002dmd: superblock update time inconsistency -- using the most recent onemd: freshest: hdl1md: kicking non-fresh hdk1 from array!md: unbind />Step 3: Hotadd missing disk [root@hassam backup]# raidhotadd /dev/md0 /dev/hdk1Step 4. Check your progress. [root@hassam backup]# more /proc/mdstat Personalities : [raid5] read_ahead 1024 sectorsEvent: 3 md0 : active raid5 hdk1[7] hdb1[0] hdc1[1] hdd1[2] hdi1[3] hdj1[4] hdl1[6] 1465175424 blocks level 5, 64k chunk, algorithm 0 [7/6] [UUUUU_U] [ recovery = 0.0% (40052/244195904) finish=405.9min speed=10013K/secunused devices: />Go get yourself some tea.. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=490676 Wed, 21 Mar 2007 19:10:00 +0100 490676 http://emelamud.blogspot.com/2007/03/blogs-going-nuts-over-impeachment.html This is truly amazing, reddit.com , the social bookmarking/news website has gone totally bananas over Bush impeachment stories. In the list of top 15 most voted up newsworthy stories, 12 are about impeachment possibility. May be it is possible.. just may be.. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485992 Thu, 08 Mar 2007 03:37:00 +0100 485992 http://emelamud.blogspot.com/2007/03/here-are-some-pictures-from-robs.html I took these pictures during farewell lunch for Rob. With few exceptions, most of them turned out quite well. .dtop,.dbottom{display:block;background: white /* (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485993 Wed, 07 Mar 2007 04:31:00 +0100 485993 http://emelamud.blogspot.com/2007/02/unlink-empty-directories-using-perl.html You have a large number of empty subdirectories and you wish to remove them from your directory. Here is one liner that will do it for you.$du --max-depth=1 -k8 ./KPTN8 ./KNDC18 ./KLRC28 ./KPNA68 ./KLRG18 ./KLRC48 ./KNS280 ./KMO8 ./KNG1$du --max-depth=1 -k | perl -ne '/^8\s+(\S+)/ `rmdir $1` 'First command 'du' finds all subdirectories with size 8kb, ie empty. The list is piped into 'perl' command , which then executes rmdir for each empty subdirectory. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485994 Fri, 23 Feb 2007 04:43:00 +0100 485994 http://emelamud.blogspot.com/2007/01/fixing-bibtex-title-lines.html Bibtex can be very annoying, especially if you are trying to get right capitalization in the title lines. In order to get the right capitalization for a string cDNA for example, one needs to change the string to '{cDNA}'. This gets ugly fast, especially if you have many such strings. The solution is to put {} around the whole title line coverting @ARTICLE{Altschul1997, author = {Altschul, S. F. and Madden, T. L. and Schaffer, A. A. and Zhang, J. and Zhang, Z. and Miller, W. and Lipman, D. J.}, title={Gapped BLAST and PSI-BLAST: a new generation of protein database search programs}, journal = {Nucleic Acids Res ... to author = {Altschul, S. F. and Madden, T. L. and Schaffer, A. A. and Zhang, J. and Zhang, Z. and Miller, W. and Lipman, D. J.}, title={{Gapped BLAST and PSI-BLAST: a new generation of protein database search programs}}, journal = {Nucleic Acids Res ... Here is a snippet of perl code to do just that #!/usr/bin/perl #fix bibtex title lines while () { $TEXT .= $_; } $TEXT =~ s/title\s*=\s*\{+(.+?)\}+/title={{$1}}/isgm; print $TEXT; (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485995 Sun, 21 Jan 2007 02:31:00 +0100 485995 http://emelamud.blogspot.com/2006/12/yet-another-night-spend-computing.html Christmas day! Thesis is done!. I feel like celerbrating Christmas, and I am not even a christian. Joy.. joy.. joy to the world. I don't think there are words to describe how this feels, years of work, years!! I feel like flying, jumping of some tall building, well may be thats not such a good idea. Best part of the whole thing, I actually enjoyed doing the thesis, partly because I discovered latex, partly because I made a lot of pretty pictures in illustrator, partly because I think my research projects are best things since sliced bread. The feeling of freedom is just amazing. Happy Holidays! (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485996 Sun, 24 Dec 2006 16:48:00 +0100 485996 http://emelamud.blogspot.com/2006/11/while-reading-comments-about-story-how.html While reading comments about story how scientology "hypnotizes" its followers, I stumbled upon a following exchange between three like minded individuals.. hehe (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485998 Thu, 30 Nov 2006 15:18:00 +0100 485998 http://emelamud.blogspot.com/2006/11/fritz-vs-kramnik_25.html This day has been in making for a long time, one flesh and blood human vs notebook with 4 processors. it can't be that hard, it is not like this thing can calculated 20 moves ahead in .1 second. This is what Kramnik said about it “Of course, this computing monster keeps getting better year by year,month by month, day by day: My opponent will be incredibly strong. ButI think I can still beat it. Whenever I can fight, I’m extremelymotivated. After all, I might be the last human being to be able todefeat this machine. My team and I will be expending all our efforts tocut this so-called artificial intelligence down to size.”You go dude! All you have to do is increase temperature in the room, so that this silicon lifeform starts to overheat and starts making all sort of random errors. Or you could just accidentally drop it on the floor, couple of times!http://www.rag.de/microsite_chess/ UPDATE: This from game 2, Fritz turn to move.. and it is checkmate for Kramnik.. Oy! (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485997 Sat, 25 Nov 2006 12:36:00 +0100 485997 http://emelamud.blogspot.com/2006/11/cost-of-ink.html This is trully amazing. HP ink costs twice as much as human blood. [ BloodInk.jpg (JPEG Image, 500x402 pixels) ] (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486007 Fri, 24 Nov 2006 00:59:00 +0100 486007 http://emelamud.blogspot.com/2006/08/me-in-brazil.html Ah... Ocean... (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486016 Tue, 22 Aug 2006 22:00:00 +0100 486016 http://emelamud.blogspot.com/2006/07/picture-of-forest.html Maryland 1 Originally uploaded by emelamud. Great news, I just puchased a digital camera from my good buddy Tom for low low price of 500$. It was a great investment, now I can post more photos. Yey! (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486015 Fri, 21 Jul 2006 04:42:00 +0100 486015 http://emelamud.blogspot.com/2006/04/test-of-netblogger-command-line-tool.html Test of netblogger command line tool. Blogging from command line is something I always wanted to do. I guess I don't really feel like leaving the comfort of my linux terminal and trusted vim editor. Blackness of my screen conforts me, and I figured there must be a program out there that would allow me to stay within limits of my X-terminal. Short visit to http://www.freshmeat.net proved itself quite useful, within minutes I found what I was looking for, a nice little perl program called "netblogger". I promptly downloaded it and tried to execute it, but unfortunately it did not run. I think that most unix apps should carry a little warning sign, "Warning!!! Trying to run this program will cause you 3 hours of lost sleep time". First, I had to install couple of perl modules for CPAN. Second, the code doesn't actually work, it refused to recognize the name of my blog. No worries, I am the world's greatest perl debugger, fix line here, fix line there, and I am off running in about 15 min. Yey! Consider this to the first of many post from command line. Now only if I could upload pictures from command line too. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486014 Tue, 11 Apr 2006 07:39:00 +0100 486014 http://emelamud.blogspot.com/2006/03/thinkpad-microphones-suck.html I am wondering if I am the only person experiencing problems with internal microphone on Thinkpads. The damn thing works fine one day and than mysteriously stops working the next. I am annoyed!! Oh that’s all I got to say about it. I think it is conspiring against me, it knows I want to make a phone call and stops working at that very moment. .damn thing!!! (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486006 Thu, 16 Mar 2006 02:48:00 +0100 486006 http://emelamud.blogspot.com/2006/11/fast-load-of-csv-type-files-into.html Here is a simple and very fast procedure for loading data into postgres database. The neat thing about this procedure, is that it uses postgres psql command line directly without need for DBI interface. All you have to do is type "psql myscript.sql", and it will* drop old table structure * create new table structure * uncompress file * filter out comment lines * parse out columns seperated by tabs * arrange them in correct order * print them out seperated by tabs * load data into table * creates indexes * and cleans up after itself! Here for example I am going to import gene2go datafile from NCBI gene database. I make an sql file like this and run psql database gene2go.sql. drop table gene_go; create table gene_go ( genome_ac int, locus_ac int, go_id varchar(255), evidence varchar(255), go_description text, pmids text ); \!gunzip -c /usr/local/db/gene/DATA/gene2go.gz | \ awk -F '\t' '{ OFS="\t"; if ( ! /#/ ) print $1,$2,$3,$ 4,$5,$6 }' /tmp/gene2go \copy gene_go from '/tmp/gene2go' DELIMITER AS '\t' NULL AS '-' \!rm /tmp/gene2go create index gene_go_idx1 on gene_go(locus_ac); (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486013 Wed, 15 Mar 2006 01:20:00 +0100 486013 http://emelamud.blogspot.com/2005/12/galaxy.html Take one pressure sensative tablet, mix it with photoshop and you get a galaxy. I love how colors flow in this picture. Done with photoshop7 and medion tablet. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486012 Sat, 10 Dec 2005 20:32:00 +0100 486012 http://emelamud.blogspot.com/2005/12/drinking-at-work-is-not-allowed.html Isn't this a great picture of me? Tom took it just before I had my fifth shot of vodka. Just kidding.. no drinking at work allowed. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486008 Sat, 10 Dec 2005 05:53:00 +0100 486008 http://emelamud.blogspot.com/2005/12/alternative-splicing-in-human.html If you don't know this yet, my PhD is on alterantive splicing in humans. I made a website and database to model spliced regions onto 3d structure of proteins. Check it out, here is the link http://as3d.umbi.umd.edu/ (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486000 Thu, 08 Dec 2005 17:06:00 +0100 486000 http://emelamud.blogspot.com/2005/12/left-foot-sand-man.html I made this guy with my left foot. I thought it might be fun to make something just using one foot. I think it turned out pretty good. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486011 Wed, 07 Dec 2005 20:02:00 +0100 486011 http://emelamud.blogspot.com/2006/11/howto-install-of-highpoint-rocketraid.html Today I got a brand new box with 1820A Controller. I was so excited I rushed into computer room, took apart first computer I could find, and pluged in the controller. Connect 7 SATA drives, rebooted the compute. Yey!!! It boots. Go threw HighPoint BIOS menus and configure RAID 10. Sweet, so far so good. First thing first boot into our trusted linux CentOS 4 and see if it detects a new device. Bad luck! Not detected! Going to have to find drivers. Ok 10 min of looking around I find no evidence of driver anywhere inlinux kernel. Very bad, time to look for drivers on the net. As luck would have there are binary drivers for RHEL4 from http://www.highpoint-tech.com/. Did I tell I hate binary drivers? Download binary drivers, followed by untar, and what is it that we see ? Oh no, images of the floppy disks? What is this 1980? Who ever uses floopy disks anymore. No problem, I seen some posting in a newsgroup about opensource drivers. I better try those first. Too bad there is no link to opensoure drivers anywhere on HighPoint website. Back to newsgroups and 10 min later, I locate this link http://www.highpoint-tech.com/BIOS + Driver/rr1820a/Linux/I download the latest greatest opensource driver. Untar it and try to compile. That didn't work. I need kernel source. Not a problem, I can do that.. type in "yum install kernel-source. And there we have it. Now we need to compile kernel cd /usr/src/linux ; make menuconfig; make; Now back to driver compilation. $cd driverdir; $make KERNELDIR=/usr/src/linux-2.6.9-11.EL/ make -C /usr/src/linux-2.6.9-11.EL/ SUBDIRS=`pwd` modulesmake[1]: Entering directory `/usr/src/linux-2.6.9-11.EL' LD [M] /root/hightpoint/hptmv.o Building modules, stage 2. MODPOST LD [M] /root/hightpoint/hptmv.ko$cp hptmv.ko /lib/modules/2.6.9-11.106.unsupported/kernel/drivers/scsi/ $makedep $modprobe hptmv Here is output from dmesg SCSI subsystem initialized RocketRAID 182x SATA Controller driver Version 1.12, Compiled Jul 26 2005 22:54:53 RR182x [0]: start channel (0) RR182x [0,0]: channel started successfully RR182x [0]: start channel (1) RR182x [0,1]: channel started successfully RR182x [0]: start channel (2) RR182x [0,2]: channel started successfully RR182x [0]: start channel (3) RR182x [0,3]: channel started successfully RR182x [0]: start channel (4) RR182x [0,4]: channel started successfully RR182x [0]: start channel (5) RR182x [0,5]: channel started successfully RR182x [0]: start channel (6) RR182x [0,6]: channel started successfully RR182x: RAID5 write-back enabled scsi0 : hptmv Vendor: RR182x Model: RAID 1/0 Array Rev: 3.00 Type: Direct-Access ANSI SCSI revision: 00 SCSI device sda: 1758216960 512-byte hdwr sectors (900207 MB) SCSI device sda: drive cache: write throughone.. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=485999 Thu, 28 Jul 2005 01:15:00 +0100 485999 http://emelamud.blogspot.com/2006/11/benchmark-of-3ware-raid-controller-7500.html My trusted raid 5 system with all the copies of all my mirrored databases has died today. All data just disappered, the controlled is fine, harddrives are fine,but data is gone. Controller card pretends like raid never existed, it wants me to initiate raid again!!! But what about my data??? Oh well, nothing to do but to start over again. Might as well learn something along the way. I found this document on 3ware website for 3ware 9000 contronllers. Lets see if we can apply the same ideas to 7500. http://www.3ware.com/LinuxWP_0701.pdf" This time I am going to be smart and use raid 10. First thing first. We create raid10 system using 3ware linux tool. $tw_cli maint createunit c0 r10 k 256 p 0:1:2:3 3ware CLI info c0 Controller: c0 Driver: 1.02.00.032 Model: 7506-4LP FW: FE7X 1.05.00.065 BIOS: BE7X 1.08.00.048 Monitor: ME7X 1.01.00.038 Serial #: F14003A4210370 PCB: Rev4 PCHIP: 1.30-66 ACHIP: 3.20# of units: 1 Unit 0: RAID 10 467.52 GB ( 980464640 blocks): OK # of ports: 4 Port 0: Maxtor 6Y250P0 Y635JKQE 233.76 GB (490234752 blocks): OK(unit 0) Port 1: Maxtor 6Y250P0 Y633J3NE 233.76 GB (490234752 blocks): OK(unit 0) Port 2: Maxtor 6Y250P0 Y637VEPE 233.76 GB (490234752 blocks): OK(unit 0) Port 3: Maxtor 6Y250P0 Y62MA44E 233.76 GB (490234752 blocks): OK(unit 0)Now let see how fast we can write to it. sync; time `dd if=/dev/zero of=/dev/sda bs=1M count=10000 sync` 10000+0 records in 10000+0 records out real 2m41.258s user 0m0.040s sys 1m10.920sSo we written 10Gb in 2min 41sec making it 61 MB/sec. Thats pathetic. I repeated the test twice getting the same numbers. Now what about read speed. sync; time `dd of=/dev/null if=/dev/sda bs=1M count=10000 sync` 10000+0 records in 10000+0 records out real 2m33.020s user 0m0.050s sys 0m44.780s Wooh.. the speed is 65 Mb/sec. I thought it would be faster for reads. Much too slow. Here is the tip for 2.4 kernel suggested by 3ware. What happens if we tune sysctl. [root@node20 root]# sysctl -w "vm.min-readahead=2048" vm.min-readahead = 2048 [root@node20 root]# sysctl -w "vm.max-readahead=2048" vm.max-readahead = 2048 [root@node20 root]# sync; time `dd of=/dev/null if=/dev/sda bs=1M count=10000 sync` 10000+0 records in 10000+0 records out real 1m45.383s user 0m0.070s sys 0m51.710s Success at last!!! I get 95 mb/sec for reads. Write speed remains the same though. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486005 Thu, 28 Jul 2005 01:14:00 +0100 486005 http://emelamud.blogspot.com/2006/11/gw-in-linux-source-code.html I find this funny.. Georege W.Bush makes into linux kernel source code. http://info.ccone.at/INFO/Mail-Archives/redhat/May-2005/msg00296.html http://lxr.linux.no/source/drivers/md/Kconfig 131 ---help--- 132 WARNING: RAID-6 is currently highly experimental. If you 133 use it, there is no guarantee whatsoever that it won't 134 destroy your data, eat your disk drives, insult your mother, 135 or re-appoint George W. Bush. (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486001 Thu, 16 Jun 2005 01:14:00 +0100 486001 http://emelamud.blogspot.com/2006/11/restarting-posgtres-server-after-disk.html Today I bravely faced a first serious recovery error with my Postgres 7.4.1 database server. I was running a large import over night and when I came back in the morning, I found my server fatally wounded after it ran out of disk space. I shut it down, freed up some space, and tried to start it again. But no luck! The server log produced the following error: LOG: checkpoint record is at 7B/66F5EFC LOG: redo record is at 7B/1080C; undo record is at 0/0; shutdown FALSE LOG: next transaction ID: 53015044; next OID: 440756341 LOG: database system was not properly shut down; automatic recovery in progress LOG: redo starts at 7B/1080C PANIC: could not access status of transaction 53119788 DETAIL: could not read from file "/usr/local/postgres/data/pg_clog/0032" at offset 172032: Success LOG: startup process (PID 30047) was terminated by signal 6 LOG: aborting startup due to startup process failure Oh this got me upset, since I have 100gb worth of data on that server, and I don't really feel like restoring them from backup. Lucky for me, I was not the first one to get this error. I found solution on Google groups. The trick is to create a missing file in clog directory. In this case, I needed to make /usr/local/postgres/data/pg_clog/0032 file. cd /usr/local/postgres/data/pg_clog/ dd if=/dev/zero of=zeros count=16 cat zero /0032 After that, I restarted the server again, and after a bit of grumbling, it finally managed to start up. Ps. This might only be a problem with 7.4.1 (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486010 Mon, 11 Apr 2005 01:16:00 +0100 486010 http://emelamud.blogspot.com/2006/11/sources-for-non-redundant-pdb-sequences.html 1. ASTRAL website / Associated with SCOP Astral Link 2. RCSB NR ftp site (blast clusters, weekly updates)3. NCBI nrPDB ( escore based clusters) URL Service (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486009 Tue, 15 Mar 2005 01:20:00 +0100 486009 http://emelamud.blogspot.com/2006/11/backup-script-for-postgres-database.html This is a simple script to create and archive for each database in the postgres server. The script first gets the list of databases from the server and runs pg_dump for each database. #!/bin/bash PSQL=/usr/local/postgres/bin/psql DUMP=/usr/local/postgres/bin/pg_dump PREFIX=`date +%j` BACKUP_DIR=/dbbackup/picasso-postgres Databases=`$PSQL -hnode0 -Upostgres -tq -d template1 -c "select datname from pg_database"` renice 20 $$ #Databases='test'; #forget above.. dump only this database echo Backup started ... for db in `echo $Databases` do echo "time: `date +%H%M%S` - Backup of $db in progress ..." $DUMP -hnode0 -Upostgres -D $db /$BACKUP_DIR/$PREFIX.$db gzip /$BACKUP_DIR/$PREFIX.$db echo "time: `date +%H%M%S` - Backup of $db finnished ..."#check if file exists if [ ! -s $BACKUP_DIR/$PREFIX.$db.gz ] then echo "backup failed for $BACKUP_DIR/$PREFIX.$db.gz"; exit; fi done echo Backup finnished ... echo "Removing old backups!"; find $BACKUP_DIR -mtime +8 -exec rm {} \; (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486004 Sat, 30 Oct 2004 01:31:00 +0100 486004 http://emelamud.blogspot.com/2006/11/r-data-frames-in-excel.html This is a simple function for automatically openning up R data frames in the excell spreadsheet. Copy this function into your workspace and type in excel(mydataframe) at the prompt. excel = function(x) { #create a temp filename tmpfilename /> #export table in comma seperated by , write.table(x, file=tmpfilename, sep=","); #run excel as a background process system( paste('"C:\\PROGRAM FILES\\MICROSOFT OFFICE\\OFFICE\\EXCEL.EXE"', tmpfilename ), wait=FALSE); #sleep 5 seconds, this is done to insure that R doesn't delete the file before excel opens it Sys.sleep(5); unlink(tmpfilename); } (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486003 Sat, 30 Oct 2004 01:13:00 +0100 486003 http://emelamud.blogspot.com/2006/11/statistician-joke-found-this-on-on.html Three statisticians went duck hunting. A duck was approaching and the first statistician shot, and missed the duck by being a foot too high. The second shot and was a foot too low. The third cried, "We hit it!" (Source: Eugene Melamud's Blog) Eugene Melamud's Blog blogs http://www.medworm.com/rss/comments.php?id=486002 Sat, 08 Feb 2003 01:02:00 +0100 486002