Tinkering with creating a bootable USB disk with which to install Feisty on Brahms. Tried on round of effort and got to a boot error.

One how-to at my disposal suggests repartitioning the stick with cfdisk /dev/sdb

That got me to this fine error FATAL ERROR: Bad primary partition 0: Partition begins after end-of-disk Press any key to exit cfdisk, which I’m going to call helpful. Something didn’t work as planned.

Meanwhile, Oates can’t chop up a PDF in Acrobat Reader, but I can, because I’ve got pdftk. Which he could get, but I digress.

I got started, but realized I wasn’t naming things the way I ought and (of course) got to wondering why I don’t know how to use regular expressions in mv. If you want to take everything named *.pdf and rename it ABBR_{originalname}.pdf … how often do you want to do that? At least once every six months, right? So it is worth spending twenty minutes looking for a regular expression syntax that will save you a half a second. Right? (Please agree with me. I don’t know how I’ll handle it if you don’t …)

[0 amanda@stillwell ~]$ for f in *.pdf; do mv "$f" "ny_${f%}";done

It is a good starting point for doing more interesting things. Fortunately, I’ve done what I needed to do and I might never come back to this wee bit of shell scripting.

I use grep a lot to backtrace through the layers of includes that make up the site I manage, and since I’ve now got that site halfway into version control, I’m finding that I’m a wee bit frustrated because grep -r recurses through every directory, including the .svn folders. So I get a lot of extra responses. I want to tell grep not to look anywhere down the .svn path.

Or, perhaps, is there another tool I ought to try?

I know, I know. I promised about the puns. I’m not even frustrated.

Project du jour: version control with dreamhost one-click installs. When you update an installation using the Dreamhost control panel, Dreamhost moves your old files from directory to directory.old, and installs the update in directory. If you want these new files in subversion, you have to scoot all the existing .svn directories back into directory.I’ve never (really) written a shell script from scratch before and this seemed like a good place to start.

With the help of an indispensable overview of advanced bash scripting published by the Linux Documentation Project, I got this far:


#!/bin/bash

path=/home/username/staging/blogs/
oldblog=someblog.old
newblog=${oldblog/.old/}

directory=$path$oldblog

for file in $(find $directory -type d -name '.svn')
do
echo mv $file ${file/$oldblog/$newblog}
done

exit 0


I find all the .svn directories and print out a close approximation of the command that I want. A close approximation. What I still need to do is strip off the last .svn so that the result will look more like:

mv /home/username/staging/blogs/someblog.old/.svn /home/username/staging/blogs/someblog.old/

Once I’ve got that down I can actually execute the command itself instead of just writing it out. I think I’m missing something syntax-wise (I know it is in the manual someplace …) and while I puzzle over it, I need to move on to other things.

Still, if you’ve got feedback or a better way (more elegant? more right?) I’m all ears.

For bonus fun, my bash environment on Dreamhost insists on writing long commands to a single line. Why? I can’t read that shit.

Two projects I’m stuck on as Friday draws to a close…

First off, importing the GG site into my new subversion repository craps out when it gets to a particular folder:


Adding /some/path/parks/index.shtml
Adding /some/path/parks/learn.html
Adding /some/path/parks/flash
svn: Valid UTF-8 data
(hex: 2e 5f 50 6c 61 74 65 6c 65 74)
followed by invalid UTF-8 sequence
(hex: 99)


flash is a directory that contains one .fla file, one .swf file and a font directory called Platelet? (yup, with a question mark. I have some ideas about why it is crapping out, but nothing solid. I’d really like me some version control.

[update:] I couldn’t post this without trying a few more tidbits. There was a hidden file (some OSX detritus) that contained an extended character. So that is solved.

Also, (perhaps I’ll fix this before I finish blogging, too) I’m puzzling over the right way to enable remote access to mysql. Along with my fancy pants version control, I’d really like to periodically make a copy of the live database. Or to periodically update my local copy of the database with newer data. Something like that. It is a one way transaction. My options, as I see it, are: mysqldump, mysqlhotcopy and replication

Replication seems excessive, but I could be missing something.

I think hot copy is all I need, but I’m now confronted with the secret truth: I’ve never setup a mysql server before. Not properly anyhow. I don’t know where the settings hide that I need to make mysql accessible over the big bad internets, via a domain name. I don’t know exactly how to do that securely (limit connections by originating IP address?).

I know that I need to change the bind-address setting, but …
* I only want to allow connections from a finite number of IP addresses, and
* I’d like to limit the users who can connect remotely.

[update]: that was actually really easy, too. As it turns out, mysql was already accepting some external connections. The rest is just a matter of GRANT syntax.

Once, these links helped.
windows clients can’t see workgroup (a post I made to a samba list) and Chapter 38. The Samba Checklist; Part V. Troubleshooting (a diagnostic walk through).

If you’re setting up a SMB network and hitting walls, start with Chapter 38, Part V.