I’ve done a lot of windows programming when I was using the system. I’ve been creating things I needed back then. Then I moved to use mostly (and then almost exclusively) linux and continued creating things I needed there. I still use linux a lot, of course, but I’ve also added another system to my everyday life (which, I must admit, I like) — Mac OS X.

I have to admit, there’s no particular need in TFTP server and client on OS X, since it has command line tools for these needs, but still sometimes it is more convenient to fire up a nice GUI, especially when it comes to the server.

Anyway, to expand my horizons, learn a bit about the platform I use, dig a bit deeper into Objective-C, etc. I have ported the well-known PumpKIN TFTP daemon to Mac. (Icons were the hardest part).

This is the very 0.0 version, but it seems to work. And feedback is appreciated if anyone cares to try.

Love,
H

Склоняя на все лады разнообразные числительные к сожительству с разнообразными существительными во всевозможных падежах, я решил что человечество вправе об этом знать, а то и присоединиться. А то и сказать мне как я обосшибся.

Ежели кому-то хочется или нужно повторить подвиг, то этот кто-то, будучи не ленивым, может даже стырить код, это не трудно.

There are some things that are awfully out of fashion, that I still do. One of those things is maintaining server with its own configuration — personality, if you please. As opposed to maintaining the army of 125000 identical machines (yes, right, the same goes for humans).

That definitely doesn’t take any enterprise grade configuration management system unless done in corporate environment, due to the lack of managers to keep confused. For tracking configuration files one can get away with widely available tools while nobody sees. In particular, I like putting files in git repository.

So I thought I’d put some snippets here for future reference and copy pasting.

First thing is add a nice alias for the root user (that guy who is in charge of breaking configuration):

alias cfgit='git --git-dir=/var/cfgit --work-tree=/'

Or even

eval $(echo "alias cfgit='git --git-dir=/var/cfgit --work-tree=/'" | tee -a /root/.bash_aliases)

(depending on the distribution). Then initialize and configure the /var/cfgit repository:

git init --bare --shared=0600 /var/cfgit
cfgit config core.logallrefupdates true
cfgit config user.name root@$(hostname)
cfgit config user.email root@klever.net
echo /\* >/var/cfgit/info/exclude

Done. Now one can add files to track before breaking the system.

cfgit add -f /etc/tobebroken/
cfgit commit -m 'stock tobebroken package configs'

cfgit add -f /usr/local/bin/mygreatscript
cfgit commit -m 'my great script before I broke it'

and so forth…

Don’t. Really. Unless you understand it.

I had an iPhoto library of some 30K pictures — some managed, some referenced and I wanted to consolidate the library — that is to copy the originals (a.k.a. masters) to the iPhoto Library for those photos that were copied with the appropriate preference off. Searching the web, looking at tools seemed to be of no help, so this is where I had to dig into iPhoto library guts and see if I can do it on my own.

It didn’t take long to find that the Database/apdb/Library.apdb file and its RKMaster table have the information in question. Thank you, Apple, for adopting the sane standard for the databases, BTW. So, all I had to do is to copy files to the Masters directory (the subdirectory structure there is pretty straightforward — %Y/%m/%d/%Y%m%d-%H%M%S in terms of strftime) and update the database.

The shell script follows (note, that I used the script on linux, where the network storage resides, it may not be compatible with other environments out of the box). What makes it a bit longer than one might expect from the description above is that it puts no more than 2000 files in one directory.

#!/bin/bash
ipl="/path/to/your/iPhoto Library.photolibrary"
fvuuid="the-uuid-of-the-volume"
fvpath="/path/to/the/volume/on/disk"

dryrun="echo"
cpln="ln"

lf="/tmp/masterslist.$$"
trap "rm $lf" EXIT
sqlite3 "$ldb" "SELECT modelId,imagePath,originalFileName FROM RKMaster WHERE fileIsReference and fileVolumeUuid='$fvuuid'" >$lf \
|| { echo "Couldn't fetch the list of images" ; exit 1 ; }
while IFS=\| read mid ip ofn ; do
 [[ -r "$fvpath/$ip" ]] || { echo "$fvpath/$ip not found" ; continue; }
 [[ -n "$idir" && -e "$mpath/$idir/$ofn" ]] && idir=''
 [[ -z "$idir" || "$((++nf))" -ge 2000 ]] && { sleep $(((nf=0)+1)) ; idir="$(date +%Y/%m/%d/%Y%m%d-%H%M%S)" ; $dryrun mkdir -p "$mpath/$idir" ; }
 $dryrun $cpln -v "$fvpath/$ip" "$mpath/$idir/$ofn" || { echo "failed to $cpln $fvpath/$ip to $mpath/$idir/$ofn" ; exit 1; }
 $dryrun sqlite3 "$ldb" "UPDATE RKMaster SET fileVolumeUuid=NULL, fileIsReference=0, imagePath='$idir/$ofn' WHERE modelId=$mid" \
 || { echo "failed to update database to reflect copy of $fvpath/$ip to $mpath/$idir/$ofn" ; exit 1; }
done <$lf

Bold parts are the parameters to be modified. To get the uuid of the source volume one needs to peek into the database. It’s the second column of the sqlite3 Database/apdb/Library.apdb "SELECT * FROM RKVolume" output.

And again. If you don’t understand, don’t do it. At any rate, backup what is there to backup. And don’t hold it against me if you fail.

Not really a breakthrough, but the amount of small changes justifies a release. The new iii now consumes less memory when calculating integrity digest, accepts all files the card is willing to send and complains better when being wronged (the latter is only important if you haven’t got it to work yet).

Love,
H