On Mon, Nov 13, 2017 at 8:50 AM, Rory McCann <r...@technomancy.org> wrote:
> On 13/11/17 01:16, Yuri Astrakhan wrote: > >> if an accepted tool already does something in a certain way, and noone is >> raising any objections to it, I think other software should follow in the >> same foot steps. >> > > ... > >> I haven't heard anyone saying that JOSM validator autofixes do a bad >> thing until this conversation. Do you think they are bad? >> > > Yes, sometimes! I looked at your fixes, saw one that didn't make sense, > (about unisex toilets) followed it to the JOSM validator, filed a bug which > seems to be fixed ( https://josm.openstreetmap.de/ticket/15536 ). > > JOSM isn't perfect, "many eyes make all bugs shallow" etc. > > Great, thanks for spotting it! I will update the Sophox task shortly as well. But my statement about "tool doing something in a certain way" is not about the specific task - as they can be buggy in every tool. I was talking about the overall JOSM autofix approach that Sophox copies and attempts to improve. I don't think your example argues against that. On the contrary, your example shows that it is much better to have these tasks standalone, with an expert oversight. Right now you have no easy way to find when users have auto-fixed the bathrooms using JOSM - there could be none, or thousands. And they are mixed together with other changes, making it nearly impossible to revert. With Sophox, you can instantly find them all, and review/revert them - simply search changsets for task_id=josm_unisex_female_male_dup. That's why I think Sophox is a much better and safer alternative to JOSM's autofixes.
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk