Rockbox mail archive

• Next message: TP Diffenbach: "Announce: id3v2 patch 706111 updated"
• Previous message: Boris Maras: "Re: minimalistic win32 rockbox sdk update"
• In reply to: idc-dragon_at_gmx.de: "Re: Player Flash"
• Next in thread: Björn Stenberg: "Re: Player Flash"
• Reply: Björn Stenberg: "Re: Player Flash"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Quoting idc-dragon_at_gmx.de:

> What happens if there is no more reason, because the unit is shut dows
> shortly after? Are the settings lost, or do you then force a spinup?

Changing setting and then shutting down without performing a disk access results in the settings changes being lost. Any disk access, as to read a directory, however, results in the sttings being saved. Merely returning from the settings menu doesn't do this, as the directory to which you return is cached, so no disk access is required to display it.

I discovered this when, to my great consternation, a patch I'd written didn't seem to be consistently saving its settings.

What I also discovered is that the settings are copied from the global struct user_settings global_settings to the global unsigned char[] config_block continuously -- at over once a second. I wonder if this copying could also be deferred until the values are actually to be writen to the disk, as config_block exists ONLY for saving settigs to disk/RTC (the "copy" to config_block involves compressing the global_settings data with various bit masking and packing).

--Tom

-- 
Archos FM has a Rockbox!

• Next message: TP Diffenbach: "Announce: id3v2 patch 706111 updated"
• Previous message: Boris Maras: "Re: minimalistic win32 rockbox sdk update"
• In reply to: idc-dragon_at_gmx.de: "Re: Player Flash"
• Next in thread: Björn Stenberg: "Re: Player Flash"
• Reply: Björn Stenberg: "Re: Player Flash"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]