On Tue, Oct 13, 2015 at 09:02:34PM +0300, Lubomir I. Ivanov wrote: > On 13 October 2015 at 20:59, Lubomir I. Ivanov <[email protected]> wrote: > > On 13 October 2015 at 20:38, Dirk Hohndel <[email protected]> wrote: > >> On Tue, Oct 13, 2015 at 10:10:08AM -0700, Dirk Hohndel wrote: > >>> > >>> > cannot open "/home/mturkia/.subsurface/hashes" > >>> > ---8<--- > >>> > > >>> > The last line is the only one related to the change to .subsurface > >>> > directory. Apparently .subsurface is not created even though it is > >>> > used for something. This just makes me wonder what else might have to > >>> > be taken into account when changing the name of default directory. Or > >>> > is it even wise to switch for old users that already have everything > >>> > under the old dir. > >>> > >>> The plot thickens - I was able to recreate this once... but I have a > >>> .subsurface directory so I guess then the hashes were written. I'll take a > >>> look if maybe the image hashing ends up failing silently if the directory > >>> isn't there or something? > >>> > >>> Thanks for helping to track this down. > >> > >> So I ran into two things that seem to be contributing to this. > >> > >> a) we seem to not always make sure that the directory into which we are > >> trying to write things exists. The easiest way around this seems to simply > >> create our default directory. I did this for Linux and Mac. Instead of > >> needlessly breaking code... Lubomir, could you add this for Windows? > >> Anton, could you add this for Android? > >> > > > > i don't think that the path creation should be per OS file i.e. in > > system_default_path_append(). should it? > > that function is more about path forming and retriaval, not path creation. > > > > once the path is retrieved in main.cpp, main.cpp or any later stage > > could create the path using the subsurface_mkdir() helper. > > also i assume this should work on Android as long as > > subsurface_mkdir() works there. > > > > if the above makes sense and if you want me to i can revert "Make sure > our default path exists (for Linux and Mac)" and send a new patch to > create the path in main.cpp?
That sounds reasonable. Yes please. /D _______________________________________________ subsurface mailing list [email protected] http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
