On January 16, 2005 17:05, Michel DÃnzer wrote: > On Sun, 2005-01-09 at 14:52 -0500, Slaven Peles wrote: > > Perhaps it would be a good idea to put in X documentation that options > > such as AGPMode, AGPFastWrite and EnablePageFlip are experimental and may > > cause unpredictable problems when nondefault values are used. > > Where did you get the idea that this is _not_ the case for at least the > majority of X server driver options? There's usually a reason why > something is an option with a certain default value.
I got the idea from man pages. For example it says: Option "AGPMode" "integer" Set AGP data transfer rate. (used only when DRI is enabled) 1 -- x1 (default) 2 -- x2 4 -- x4 English is not my first language, but I understood this as I actually _should_ set AGP mode to x4 if this is the transfer rate my chipset supports. The sentence is written in imperative form. My understanding was that the default value was set to x1 only to include older chipsets which do not support higher rates. Besides, there is a lot of unofficial documentation on the net which actually recommends users to enable these options, and since there were no explicit warnings in official XFree86 docs I concluded these optios were safe. Furthermore, when I set AGP mode to 4 I experienced problems that didn't look like X problems at all. I suspected IRQ conflict, proper kernel modules not loaded, hotplug bugs etc. So, I think that perhaps mentioning somewhere in documentation that changing default options may cause very weird system behavior may be helpful. This is just my opinion, and you may not agree with me. My primary reason for filing this bug report is -- as I mentioned it before -- to have this problem documented somewhere in hope that it will be useful to somebody. Respectfully, Slaven