Hi Willy. Thank you for your suggestions.
Yes, We confirm that the data file is loaded only once, upon startup. In case that will change in some future release, we'll take care to document the impact with chroot. Regards -Scientiamoblie On Fri, Nov 4, 2016 at 4:43 PM, Willy Tarreau <[email protected]> wrote: > Hi, > > On Fri, Nov 04, 2016 at 11:35:05AM +0100, Scientiamobile wrote: > > Hi All, > > > > This patch will make Scientiamobile WURFL device detection features > > available in HAProxy > > > > It applies to version 1.7 > [...] > > Thank you. This patch is particularly clean. I'm seeing only very minor > stuff > that I can easily deal with during the merge (eg: configuration.txt lines > well > above the 80-chars limit). I appreciate a lot that you've put an end to the > README pollution, and I think that after merging this patch I'll move > DeviceAtlas and 51Degree's howtos out of the readme to their own file like > you did so that the readme becomes an entry point again. > > I'll take care of it next week so that we can have it in the next (and > hopefully > last) dev release. In the mean time if anyone has any comment, this is the > right > moment. > > Among the small things I'm seeing are : > typedef char *(*PROP_CALLBACK_FUNC)(wurfl_handle wHandle, > wurfl_device_handle dHandle) > and > typedef struct ... wurfl_data_t > > That don't seem to be needed in wurfl.h since other parts of the code will > not > need them. In general we *try* to limit exposure of internals, and we > almost > don't use typedefs in the code, so if we can isolate them in wurfl.c it > will > be better (I'll try it, this is easy). > > By the way, can you confirm that the data file is loaded only once upon > startup > and not touched later ? If this is not the case, you may get into trouble > when > using chroot, in which case it might be desirable to document it. > > Thanks, > Willy >
