I'll just leave this here in case it might help somebody else some day. Some of these items are copied from my old development blog I started on the Radiant Mercury programme at Lockheed Martin.
If your web server host allows only SFTP (a good idea for security), you'll find that Composer doesn't support SFTP. For reasons of tool familiarity and process, I did not want to impose a change of HTML editor on the maintainer of a certain web site, so I came up with the following workaround for users who prefer to use the Composer HTML editor and didn't want to learn a new one:
-
Start a local FTP server on the user's machine. In this case it was Mac OS X 10.5; setting up an FTP server was as simple as turning on File Sharing in the Sharing section of System Preferences, checking the firewall settings to make sure the FTP server was visible only to the local machine, and editing
/etc/ftpchrootto contain a single asterisk character for security. -
Create a directory in the user's home directory called
~/proxy_for_website/ -
Inside that directory, mimic the directory structure that exists already at the web hosting company, e.g.,
~/public_htmland~/secure_htmland all sub-directories that might exist below them. -
Inside Sea Monkey Composer, change the Publish As... settings as follows:
-
Publishing address:
ftp//name_of_user's_machine.local/proxy_for_website/public_html/ -
User name: the user's username on the local machine.
-
Password: the user's password on the local machine.
-
Now set up an SSH public key pair on the user's local machine and on the web hosting company's server in the usual way so that SSH can be done without a password (hint: use
ssh-agentto do it securely). This will ensure thatrsynccan run automatically. -
Set up a cron job on the user's machine to run the following script once every minute:
That's it.
#!/bin/sh
#
# This script watches for any change in the $target directory, and if it
# sees a change, copies whatever changed to the user's public_html directory
# at Hurricane Electric. Run this script once a minute from crontab.
start_time=`date +%s`
old_detectfile=/Users/username/detect_mozilla_upload_content.new
new_detectfile=/Users/username/detect_mozilla_upload_content.old
reportfile=/Users/username/detect_mozilla_upload_report
target=/Users/username/proxy_for_website/public_html/
rm -f $reportfile; touch $reportfile
cd $target
find . -type f -ls > $new_detectfile
diff -q $new_detectfile $old_detectfile
RC=$?
if [ $RC -eq 0 ]; then
echo "nothing to do at `date`" >> $reportfile
else
echo "\nchange in the proxy directory detected (diff rc was $RC) ; running rsync now\n" \
>> $reportfile
rsync -avz . [email protected]:public_html >> $reportfile
RC=$?
echo "\nRC from rsync was $RC" >> $reportfile
mv $new_detectfile $old_detectfile
end_time=`date +%s`
elapsed_time=$(($end_time - $start_time))
echo "\nElapsed time $elapsed_time seconds." >> $reportfile
fi
As a bonus, your user will be pleased to discover that publishing in Composer is almost instantaneous now.
Always specify the encoding! When writing HTML (especially auto-generated HTML) you should always include the Unicode encoding information in your header, like this:
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<title>
...
<title>
</head>
</html>
The <meta> tag containing "Content-Type" information must be the very first element
in the <head> block, because that's where HTML-aware applications go looking for it.
If you don't do this in your HTML, someday it will come back and bite you.
To pipe stderr along with stdout, do:
% cmd1 |& cmd2
To redirect stderr by itself to file f1 and stdout to file f2, do:
% (cmd > f1) >& f2,
Source: Daniel Gilley. UNIX in a Nutshell second edition. Sebastopol, California: O'Reilly & Associates, 1994.
If you have Ghostscript installed, do this:
% set path=d:\work_in_progress\tools\gs\gs7.04\lib;d:\work_in_\
progress\tools\gs\gs7.04\bin;%PATH%
% gswin32.exe -sPAPERSIZE=letter -dNOPAUSE -dBATCH \
-sDEVICE=pdfwrite -sOutputFile=result.pdf file1.pdf [file2.pdf...]
pfiles pid will give a list of file descriptors currently open by process ID pid.
Choose the file descriptor you want and look for the ino: nnnnn field. That is the
inode of the file. To get the name of the file, do % find . -inum nnnnn, or
% ls -i | grep nnnnn /usr/proc/bin/pfiles
This error from GitHub for Windows is caused by low disk space on the local volume containing the repository. Less than 4.9 gigabytes free: the error occurs. Freeing up some space makes the problem go away.
`:g/\%x94/s//"/g
% Tell Adobe Acrobat Reader not to display the bookmarks pane.
\hypersetup{pdfpagemode=UseNone}
% Tell it to start in my preferred size and skip front matter.
\hypersetup{pdfstartview=FitH}
\hypersetup{pdfstartpage=12}
% Fill in some other information in the PDF header.
\hypersetup{pdftitle=Security Test and Evaluation of Cross Domain Systems}
\hypersetup{pdfauthor=Joe Loughry}
\hypersetup{pdfsubject=computer security}
\hypersetup{pdfkeywords={doctoral thesis}}
Use mkfile
% mkfile -nv 20g 20gig_file
It's not fast, though. The available disk space goes down immediately, but the command hangs for a
long time writing zeros to the file. Stackoverflow suggests a faster method using dd and seek:
% dd if=/dev/zero of=filename bs=1 count=0 seek=200G
The dd method is no faster, though, and it's harder to interrupt with Ctrl-C.
Hurricane Electric, a great web hosting and co-location company, recently
started disabling .virtualmail wildcarding functionality, which I depended on for ad hoc email
addresses. The workaround is simple, because they allow me to create new mailboxes, and set
forwarding on them. Forwarding can even be done to accounts on other systems, although doing that
is discouraged because it attracts spammers.
I created a bunch of new mailboxes that forward to my real mailbox:
- mailto:[email protected] forwards to mailto:[email protected]
- mailto:[email protected] forwards to mailto:[email protected]
- mailto:[email protected] forwards to mailto:[email protected]
- mailto:[email protected] forwards to mailto:[email protected]
Messages sent to any of these addresses show up in one place, and their 'From:' address is preserved so I can tell where it came from and notify the sender that they need to update their address books.
To see more information in Mac OS X, sudo nvram boot-args="-v" in Terminal.
rsync -cir local_directory/ user@remote_host:remote_directory/
The -c option tells it to compare checksums instead of date/time; -i itemises
the changes it makes, and -r recurses into sub-directories.
Use ipfw to throttle a particular protocol, e.g., a Bitcoin Core client
operating on 8333/TCP, like this:
#/bin/sh
ipfw_cmd=/sbin/ipfw
#
# This function checks to see if the script is running with root privs.
#
function are_we_running_as_root_interrogative
{
if [[ $EUID -ne 0 ]]; then
echo "Run this script as:"
echo " $ sudo ./$0"
exit 1
else
echo "The script is running as root."
fi
}
are_we_running_as_root_interrogative
$ipfw_cmd -f flush
# Throttle only connections made outgoing by Bitcoin Core.
$ipfw_cmd pipe 1 config bw 200Kbit/s
$ipfw_cmd add 50 pipe 1 dst-port 8333
$ipfw_cmd pipe 1 show
To get the process ID of a running process in Bash, it is necessary to
pre-process the output from ps before parsing it with cut, like this:
/bin/echo -n "Starting Tor..."
tor --quiet -f $torrcLoc &
pid=`ps ax | grep "[t]or --quiet -f" | tr -s " " | cut -d ' ' -f 2`
if [[ $pid -ne 0 ]]; then
echo "the process was started with PID $pid."
else
echo "a problem occurred."
fi
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent