Keith's Personal Blog

OK, So i've been forced to have to set up an eggdrop -- what a ball ache!

Both ubuntu and Debian are now shipping a threaded version of TCL, unfortunately this mean eggdrop doesn't work, as for some reason it fails to detect properly if tcl is compiled as threaded.

It appears to start, but won't connect to a server properly, and won't allow you to telnet to it either.

Appears the easiest solution (from http://eggwiki.org/Threaded_Tcl) is after running ./configure. edit config.h and find #undef HAVE_TCL_THREADS, and change it to #define HAVE_TCL_THREADS 1

If you have allready compiled the eggdrop once, run make clean to clean out the compiled stuff, and then compile as normal (make && make install), and that should fix it, at least it has for me!

Not sure if/when the configure script is going to be fixed tho.

Keith

If you've ever wanted to clear the contents of a file quickly, without breaking anything hooked into a file (useful for large files), look no further than the truncate command provided by the coreutils package.
Instantly Shrinks the file to 0 bytes, and maintains anything associated with it.

Just be careful you don't accidently clear a file by accident, as theres no way to reverse, and it doesn't prompt you to confirm your selection

Keith

OK, so for anyone who uses OSX , and admins Linux machines, this can be a real headache.

For some reason the default settings for your Home and End keys that come with Terminal.App for OSX don't do what you expect - theyre Actually Swapped with Shift Home , and Shift End, but thats not all, as the escape sequence for Shift Home for example, is set to "\033[H". This seems to work fine on OSX, but unfortunately you try logging into a Linux Box with that, and on the command line it works, but try loading vi and using the Home key - and you get nothing. The same applies to the End key as well, OSX has it defined it as \033[F, however vim remotely does not like this either. (I Believe this applies to all termcap applications, allthough i can't be sure).

The Solution for both of these is to change the definitions of these keys, so theyre mapped as you expect, and get it to send the right escape sequence at the same time, I don't believe mapping OSX's [H and [F mappings in vim on the remote host, because well , when you admin 10 machines or more, it well doesn't make sense to put it mildly, so this is the only viable solution.

Before we can do this however we need to create a file, open up a Terminal window, and type the following: sudo nano /private/etc/inputrc, then enter your password (you need to be an admin user for this). Now paste in the following code: followed by Pressing CTRL and X, then typing y, and hitting enter.
(Note: If you don't want to do this change system wide, change the command to nano ~/.inputrc, and add export INPUTRC="~/.inputrc" to your .profile)

Now we've done that we shall deal with Home/End, hit CMD I in terminal, and click Settings, followed by double clicking whichever entry is selected. In the window that appears click on the "Keyboard" tab, and find the entry for Home, click on it , and click edit. Change the action to "send string to shell:", now in the box below that, hit the following keys: "[1~" ( as in the escape key, and without the quotes) and click OK. Now do the same for End, except you type "[4~" (without the quotes).
Next find Shift Home, and edit that to change its action to: "scroll to start of buffer" then do the same for Shift End, editing that and changing its action to: "scroll to end of buffer".

Right, now Restart Terminal.

Try your home and end keys, they should work, both on command line and inside vim. Both locally and remotely. If they don't try hitting CTRL V .. it should reply with ^[[1~ .. and CTRL V should reply with ^[[4~. If it doesn't, you've gone wrong somewhere!.

Comments welcome.

Keith

OK, so this is just a heads up for anyone who knows me, due in part to my poor health these days i've withdrew from a few projects i was participating on.

Unfortunately this seems the only way I can maintain working on other things.

Apologies to anyone who may have been waiting for me to develop something, and I have not provided. I also apologise for my lack of activity, unfortunately at the moment it cant be helped.

Just so people are aware, I have also chosen to drop ircu (undernet-ircu) development, with the current release of ircux-2.10.11.08. Instead i'll be concentrating development efforts towards gnuworld, and creating a port of it that supports more than just postgresql, (Mysql Support is allready complete), and supports connecting to TS6 networks (namely inspircd). Along with modularising some of the features gnuworld provides, so further expansion will be possible.

As some of you are aware, I am the primary developer of the Netgamers Gallery, development on Version 2 of the site (The one Currently online) is limited to bug fixes only, and the receipt of feature requests is welcome for the next version (v3) , I hope to get this completed and released as soon as possible, but cannot give a release date due to my afformentioned health problems.

Planning on Marvin (A Modular IRC Bot in C++) is going well, the svn repo for it has been created, and initialised, and will have more information about its featureset soon.

Unfortunately these are the only projects I shall be maintaining for at least the near future. I may re-evaluate this over time however. If there is a project or problem you think i should be working on, and its not listed here, please drop me an email via archerATpriorityonlineDOTnet or keithATchaos-realmDOTnet.

Until Next Time!

Keith

OK , it seems some times when installing packages, the installation scripts get a bit confused, and can't seem to handle initscript errors.

This results in said package being half installed, but marked as failed, On occasion running apt-get -f install or dpkg --configure -a, once you have solved the problem will get things into working order again.

These commands don't allways work however, and your left with a package stuck like this:

user@server:~$ sudo dpkg -l lighttpd
iF lighttpd 1.4.25-2 A fast webserver with minimal memory footpri

(iF meaning installation Failed)

In order to resolve this, if it still refuses to install properly (even after fixing the problem that caused the error in the first place), it is neccessary to force remove the package (unfortunately dpkg --purge or --remove on its own will not work).
Typing the following (replacing lighttpd with your package name), should get the system back into working order again.


Then if you try running apt-get -f install, it should bring your system back into the correct state again.

There are other ways around this problem (I think manually installing the .dpkg archive via dpkg -i may work, but i havent tried it), but i think this is the easiest.

Anyways Tara for now!.

Keith

OK, it appears as though somone in the debian community thought it was a good idea to make apt-get pull in Recomended packages, when it looks for dependencies, so therefore, when you try to install something like openssh-server for example, it grabs all the Recommends as well, which happens to include X11!.

Its easy to turn this off however by adding the following:


to either /etc/apt/apt.conf or by creating a file inside /etc/apt/apt.conf.d/ like 00NoAptRecommends.

This at least for me, stops this most unwanted change... why they actually did this i shall never know.

[ Credits goto: blogs.KoolWal.net for finding out how to turn this off!]

Keith

Right ,

I've recently had the pleasure of playing with mod_ruid again, and I've come across an issue with how it functions and mod_cgid,

Normally the cgid unix socket is owned by the userid of the parent process, but because ruid calls setuid it can no longer access the socket due to the permissions, meaning all calls to cgi scripts fail.

I've therefore had to switch out mod_cgid in favour of mod_cgi.

I'm unsure of any other way around this, as changing the permissions on the mod_cgid socket doesn't sound like an ideal solution,

Oh well, running mod_cgi seems to work sufficiently for my needs.

OK , so i've gotten around to actually setting KDE as my wm for X11 on OSX.

One of the things i had to do, in order to load any kind of fink based app was to create .bashrc inside my home directory directory with the following contents "source .profile" because bash doesn't by default look inside .profile (where your fink init line is) unless its actually told to.


Keith

Well , what can i say, i've never come across a piece of software thats more awkward to configure.
It appears as though you need to learn a whole new language, just in order to enable something simple like SMTP_AUTH.
I'm having to change my plans and stick with postfix for the moment, especially considering my time constraints.
I may try exim again in future, but i think for now its postfix all the way, especially as it seems to do everything i need of it at present.

Sigh

No Clue why, but i absolutely love this track.

I'd never heard it before i got ahold of the Peter Gabriel Hit album, but its got this excellent mix of instruments, and a catchy tune to boot.

Ah well , back to work i suppose :/

Keith

OK,

So it appears we have a PNG that only appears on OSX Based Browsers (So-Far).

Every other browser, either says the image has erors, or just doesn't display it.

Try it yourself: http://files.leymoo.com/irc/vectory.png

Stranger still, any time somone gets the image, i get a "Broken pipe: core_output_filter: writing data to the network" if they can't view it. Its just odd.
UPDATE:
earthian seems to have found it as this:

22:43:14 < earthian> vectory.png is a 576x720 MacPaint image
22:43:14 < earthian> Default gamma is arbitrary for bitmap
22:43:14 < earthian> Using DirectColor visual
22:43:14 < earthian> Building XImage...done

... but WTF?

Keith

Been thinking today about yet another idea i had..

Basic Idea is this: Have a daemon similar to an IRCd, which allows other servers to link to it, but does not broadcast the servers to any of its links, unless explicitly configured to.

The basic principle is that you could link a set of services (or multiple) to it, and that all Servers could see the services, but thats all, i suppose it'd be useful for those small nets who can't afford to run there own services, and resort to stuff which doesn't fit there needs, this way services could service multiple servers, but without advertising other users.

Of course, the services would have to be modified in order to not leak information about other servers to the link its communicating with, but its a curious concept, that i may do something about at some point :/

Then again it could be made to create a meshed network .. allow the servers to communicate via it, and if there is any conflicts between the 2 links , e.g. clients appearing twice, ignore the conflicts.

Its a curious concept, and I'll probably work on something to do with it at some point in the future , but sadly my workload sucks right now

Ah Well, Back to work.

OK So it seems that if you have both IPV4 and IPV6 transit, for some reason SSH prefers IPV6 transit, over IPV4.

Im not sure why this is, perhaps its because IPV6 is meant to supercede IPV4, whatever the reason, this causes lag when using SSH, primarily due to the lack of capable IPV6 upstream providers, until this situation improves, there's a simple change you can do in order to force ssh to use IPV4 first!.

Add the following to ssh_config (/etc/ssh/ssh_config or /etc/ssh_config).

AddressFamily inet

This forces ssh to use IPV4 by default instead of v6, whilst maintaining the ability to connect to ssh capable ipv6 hosts.

Note: Any preference over v4/v6 specified in your ssh command (e.g. ssh -6 user@host) will override this setting.

Enjoy.

OK , as anyone who uses irssi knows, its irritatingly easy to accidently type /win into a channel or miss the / off alltogether, i've created a perl script that detects when you have done this and switches to the window (if it exists), this may not be the best way of doing this, but its the quickest way to solve a major annoyance for now, FYI, its based upon the original oops.pl script thats available from scripts.irssi.org, I've just added a section to deal with /win mistakes.

Any comments on this would be appreciated though.

CLICK HERE

Regards,

Keith

For all those who know me, it appears that im an idiot - I've managed to break a bone in my foot (if u really want to know how, i suggest u ask me on IRC), so it means im stuck in one place now - fs.

Ah well, suppose its more time for IRC Projects (FUN).