On 25 July 2016 at 20:51, Attila Tajti <[email protected]> wrote: > The fairly straightforward change in corpus.go (Corpus.PermanodeLatLong) in > https://camlistore-review.googlesource.com/6456 would enable geotagging > using permanode attributes "latitude" and "longitude", i.e. they would show > up in location searches, but would not filter photos having another location > in Exif. > > The other bits of https://camlistore-review.googlesource.com/6456 are about > returning the Exif location info already in the index. > > I don't know if location info is so important that it warrants special > handling.
I think location is a pretty damn important info for humans, and therefore we do need to have a way to make it easily searchable/described to clients. And to let whatever's in a file (EXIF GPS) be overriden by whatever is set in the containing permanode (if the info is more recent, as usual). > Perhaps it would be better to return everything (or just selected > fields) from Exif instead. Yes, we could do as you were saying for the audio attrs, and index a lot of the EXIF that way, but I think we should do that in addition to the above (and probably later on). > That would complicate location attribute lookup a > bit for clients like the map UI from Eric, because clients would need check > first the permanode attrs, and if they are missing also try the file > tags/attrs (once they are returned in DescribedBlob). I don't think that's such a terrible idea to let the clients parse the described blobs and decide where they want to get their info from. That's kindof what the webUI does when e.g. it chooses what title it should show for an item (filename if exists VS title attr on the permanode VS blobRef if all fails). imho, the key is what we decide to do with Location *camtypes.LocationInfo in DescribedBlob, and how we define camtypes.LocationInfo. The rest should follow from there. See corresponding comment in CL please. > On 25.07.2016 20:25:56, Mathieu Lonjaret <[email protected]> wrote: > > Thanks for the reminder about that issue. > > Given that: > 1) the search pkg already supports looking for a location, and that it > looks for either a permanode based location, or a file based location > 2) there currently is no way for a user to effectively set a location > on a permanode (because the indexer will ignore it) > > then I think the most urgent item is to let a user set a (GPS > coordinates) location on a permanode as simply as possible, isn't it? > > I may be missing a point that was made in > https://github.com/camlistore/camlistore/issues/778 against, and if > so, please point me to it. > > Attila, if you agree with the above, do you think > https://camlistore-review.googlesource.com/6456 could become the > implementation for it? In any case, I'll have another look at it > tonight. > > > On 24 July 2016 at 22:49, Attila Tajti wrote: >> After being stuck with the issues mentioned by Matthieu and seeing the >> problem with time values in Exif spec itself (no timezone) I decided to >> experiment with file metadata formats instead. >> >> In the end I started my own Exif package which can also write exif >> metadata, >> not just read it like github.com/rwcarlsen/goexif does. With it one could >> embed GPS location in the jpeg file, perhaps based on location attributes >> of >> its permanode in camlistore. >> >> The function indexEXIF pkg/index/receive.go stores all Exif tags in their >> original, close to raw format. Perhaps some fields should have special >> processing: >> - GPSLatitude/Longitude in float instead of rational array >> - GPSTimeStamp (rational) recorded together with GPSDateStamp in RFC3339 >> - time stamps (DateTimeOriginal and DateTimeDigitized) should use RFC3339 >> (without TZ when missing from exif) >> >> These could then be returned similar to media tags in search.Describe. One >> problem is that there are at least 117 exif tags I found in the spec (see >> github.com/tajtiattila/metadata/blob/master/exif/exiftag/name.csv). >> Perhaps >> we should just store (and implement query and describe for) selected tags >> instead, like it is done for audio files. >> >> >> On Wednesday, July 13, 2016 at 12:23:45 PM UTC+2, mpl wrote: >>> >>> I believe this is pretty much >>> https://github.com/camlistore/camlistore/issues/778 and related to >>> https://github.com/camlistore/camlistore/issues/777 >>> >>> >>> On 12 July 2016 at 16:44, wrote: >>> > I understand (and see) how photos with GPS data are parsed and you can >>> > search for location... no problem there. >>> > >>> > What about photos that don't have GPS data? I could add it via EXIF >>> > before >>> > import, but I would think it could be nice to add as a parameter (that >>> > is >>> > then recognized) inside of Camlistore. >>> > >>> > Actually, I would (ideally) be able to assign locations to things >>> > beside >>> > pictures (files, notes, whatever) and search them the same way... >>> > >>> > Not implemented, or am I missing the obvious? >>> > >>> > :-) >>> > >>> > Thanks all... >>> > >>> > -- >>> > You received this message because you are subscribed to the Google >>> > Groups >>> > "Camlistore" group. >>> > To unsubscribe from this group and stop receiving emails from it, send >>> > an >>> > email to [email protected]. >>> > For more options, visit https://groups.google.com/d/optout. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Camlistore" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to a topic in the > Google Groups "Camlistore" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/camlistore/QKT1kpnvGoo/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "Camlistore" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Camlistore" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
