Author Topic: Dmenu 0.6 Released  (Read 24659 times)

massivebasset

  • Guest
Re: Dmenu 0.6 Released
« Reply #60 on: December 20, 2009, 04:54:11 pm »
@massivebasset
I have also run into the disk corruption, but I'm not sure what the cause is.  I know Booboo says he thinks he has resolved the issue with dingux corrupting the SD card, but I am finding it hard to believe.  There is nothing that is done in Dmenu that is outside the normal realm of a standard C program.  The only addition to dmenu 0.6 was adding fsync'ing to the file writes, which may ultimately be the culprit.  Again, calling fsync on file writes is a common/encouraged practice, but I think there is still something wrong Dingux.    I have a ticket in the project to enable read only mode via the dmenu.ini file.  The code is in place, I just need to read that config file.  Plus with a new startup/shutdown sequence the need for writing to the filesystem decreases.  I very well may try to remove the fsync today and see if that makes any difference in the time between usefulness to disk corruption.  If you have any other information on what you are doing before corruption, let me know, otherwise I will just try it myself.

Thanks for the note.  Ill keep an eye on it and post back here if I notice it happening again.  I dont mean to claim that Dmenu is the culprit, but its the most noticeable when the exit screen no longer changes.  I reformatted to 32kb blocks, and so far have been pretty stable.

SavedLinuXgeeK

  • Guest
Re: Dmenu 0.6 Released
« Reply #61 on: December 20, 2009, 08:48:04 pm »
@all
With respect to the desire for a loading screen, work is in place to support the new loading screen, but I need a good image to use (and I don't want to rely upon the user's background image).  I set up a forum thread for a request for a background image and better menu sounds.  Please feel free to visit the forum post and give your input.  http://boards.dingoonity.org/dingux-general/dmenu-enhancements/

@massivebasset
So I played with removing fsync from the code base and it seems to be more stable now. Its a compile time flag, so that whenever dingux can handle the writes better, we can turn it back on, but it should behave better in the next release.

 

Post a new topic