I installed the Firemacs Firefox add-on for making Firefox respond to Emacs shortcuts, but I ended up removing it: it seems that Iâ€™ve spent so long down-translating my fingers on non-Emacs applications like Firefox that itâ€™s easier to continue that way than to change to using the keyboard properly (or â€œlike god intendedâ€ as Norman Walsh put it at http://norman.walsh.name/2006/11/08/firefox20). Mind you, Iâ€™ll probably enable it again after the next time that I use C-n in a text box in the browser.
When I started using ECB (http://ecb.sf.net) with the Ubuntu Emacs 22 snapshot (both X and GTK variants), the CPU usage soon shot up to 100% and stayed there.
The solution, posted at http://sourceforge.net/mailarchive/forum.php?thread_id=30578186&forum_id=1620, was to update the installed version of semantic-idle.el because of how Emacs 22 uses timers.
Unfortunately, I couldnâ€™t find an easy way to persuade Emacs to load a version of CEDET built from CVS instead of from the Ubuntu Debian packages.
I spent far more time last week making two monitors not work than I did making them work, so hereâ€™s I few things I have learnt:
- Donâ€™t use the â€œnvâ€ driver. I couldnâ€™t get it to drive two screens at once, much as I would have liked it to.
- Donâ€™t use â€˜Option â€œConnectedMonitorâ€ â€œdfpâ€â€˜ with the â€œnvidiaâ€ driver. I did RTFM rather too well and put that in as a hint to the driver. While ever it was in the xorg.conf file, only one screen at a time would work.
The eventual solution was, surprise, surprise, to download the latest version of the Nvidia driver â€” which was released partway through this epic battle â€” and follow its instructions to the letter â€” including removing existing Ubuntu packages as directed. I donâ€™t know whether the driver had improved or my technique had improved by the time I tried that third version, but it worked for me.
Of course, the first thing I did when both screens worked was see how many 80-column Emacs frames would fill the screens: over 600 lines of Emacs visible at once, now that makes it all worthwhile!
I started out using the two monitors as separate X screens but Iâ€™m now using Twinview so itâ€™s easy to Alt-Tab amongst the Emacs frames. The X startup log shows that X thinks the combined screen is too wide to be allowed, so it starts at a lower resolution, but once Iâ€™ve logged in, I can jigger the â€œNVIDIA X Server Settingsâ€ tool to get it back to itâ€™s highest resolution.
In keeping with the long tradition of filling the Internet with example configuration files that are all purported to work but are all somehow puzzlingly different, hereâ€™s the xorg.conf file thatâ€™s working for me:
# nvidia-settings: X configuration file generated by nvidia-settings
# nvidia-settings: version 1.0 (buildmeister@builder26) Mon Feb 26 23:38:28 PST 2007
Screen 0 â€œScreen0â€³ 0 0
InputDevice â€œKeyboard0â€³ â€œCoreKeyboardâ€
InputDevice â€œMouse0â€³ â€œCorePointerâ€
Option â€œXineramaâ€ â€œ0â€³
# generated from default
Option â€œProtocolâ€ â€œautoâ€
Option â€œDeviceâ€ â€œ/dev/psauxâ€
Option â€œEmulate3Buttonsâ€ â€œnoâ€
Option â€œZAxisMappingâ€ â€œ4 5â€³
# generated from default
# HorizSync source: edid, VertRefresh source: edid
ModelName â€œApple Cinema HDâ€
HorizSync 49.3 – 98.5
VendorName â€œNVIDIA Corporationâ€
BoardName â€œQuadro FX 1500â€³
Option â€œTwinViewâ€ â€œ1â€³
Option â€œmetamodesâ€ â€œDFP-0: 2560Ã—1600 +2560+0, DFP-1: 2560Ã—1600 +0+0; DFP-0: 1280Ã—800 +1280+0, DFP-1: 1280Ã—800 +0+0â€³
Modes â€œ1600Ã—1200â€³ â€œ1280Ã—1024â€³ â€œ1024Ã—768â€³ â€œ800Ã—600â€³ â€œ640Ã—480â€³
I now have the static build working after battering my head against this for a couple of days. The most cogent advice that I found was post #6 in http://forum.skype.com/index.php?showtopic=76815, which was to use â€œdpkg -xâ€ and copy the extracted libraries to /usr/lib32.