Work, play, and everything in-between.
At work, I do a lot of directory traversal: going from one location to another within the whole (global) filesystem structure. So, to conveniently go back to a previous directory, I use pushd and popd, aside from the usual cd:
$ pwd /home/iandexter $ pushd /etc/sysconfig $ pwd /etc/sysconfig $ popd $ pwd /home/iandexter
You can even echo $DIRSTACK to list the current directories in the stack, and push multiple directories.
I’ve just started using Twitter, an up-and-coming web service that posts user status on the web, in IM and in SMS. The service also exposes its API, so it’s possible to build apps using it.
I found a nifty way of updating my Twitter status through the command line:
curl -u username:password -d status='status_goes_here' -s http://twitter.com/statuses/update.xml > /dev/null 2>&1
Hmmm…
In Bash, I would have done:
$ for i in `seq 100` ; do if [ `expr $i % 15` -eq 0 ] ; then echo FizzBuzz; elif [ `expr $i % 3` -eq 0 ]; then echo Fizz; elif [ `expr $i % 5` -eq 0 ]; then echo Buzz; else echo $i; fi; done
Then again, it took me about 10 minutes to do that. (I couldn’t get the ternary operator to work, somehow.) In C: about five minutes (rusty — I last used C way back in college {ages ago!}).
In a recent phone interview, I was asked to programmatically (in shell) rename a set of files. I blundered for about two minutes, and gave up in the end, saying I could probably do that by experimentation. I could fairly say I passed that interview. The point? Answering “FizzBuzz” questions does not reflect real-world situations — it’s how you approached the problem, even if you didn’t get the answer, that matters.
If I were an interviewer, I’d concentrate on the steps rather than the solution.
I know there are scripts out there that does full-blown SMTP testing. I’ve used smtp-sink and smtp-source in Postfix for this purpose:
$ smtp-source -s 100 -m 100 -f sender@domain.com -t \\
recipient@domain.com server.address:25
So this was what I recommended to a friend, a Unix engineer in Singapore. But she had a different requirement: she has to establish an SMTP connection through telnet to a remote server, and send a template mail to thousands of recipients. The telnet session goes something like:
[user@host ~]$ telnet remote.host 25 Trying 123.456.789.10... Connected to remotehost.remotedomain (123.456.789.10). Escape character is '^]'. 220 mail.remote.host ESMTP Postfix EHLO some.domain 250-mail.remote.host 250-PIPELINING 250-SIZE 10240000 250-VRFY 250-ETRN 250 8BITMIME MAIL FROM:sender@this.domain 250 Ok RCPT TO:recipient@another.domain 250 Ok DATA 354 End data with >CR<>LF<.>CR<>LF< Subject: This is a test From: Sender This is only a test. . 250 Ok: queued as D81D7FA927 QUIT 221 Bye Connection closed by foreign host. [user@host ~]$
With thousands of recipients (no, her employer is not a spammer), this can be quite tedious. I thought of using an alias file at first, but the requirement was SMTP through telnet. So I told her to do the following:
recipient1@some.domain recipient2@another.domain ... recipientX@yet.another.domain
Subject: This is a test From: Sender This is only a test.
#!/bin/bash SENDER=sender@this.domain (echo "HELO some.domain"; sleep 1; echo "MAIL FROM:" $SENDER; sleep 1; for i in `cat /tmp/recipients`; do echo "RCPT TO:" $i; sleep 1; done; echo "DATA"; sleep 1; cat /tmp/message; sleep 1; echo "."; sleep 1; echo "QUIT") | telnet remote.host 25
chmod 744 smtp-telnet.sh and run the script:
$ ./smtp-telnet.sh
The sleep parameter can be increased to compensate for delays. I tested it on 50 users in my local test environment, and it worked for me.
