Re: Endless gamma in eXIf 2017-0119
John Bowler <[email protected]>
| Newsgroups | gmane.comp.graphics.png.general |
|---|---|
| Message-ID | <CAP7U399c=LZ+fsL9C9b2A78bEqwpGp9X54tNOJ+tE0ytt8ehVA@mail.gmail.com> |
On Tue, Feb 7, 2017 at 10:38 AM, Glenn Randers-Pehrson <[email protected]> wrote: > It's not sufficient because our copy-safe mechanism does > not adequatly handle potential conflicts among ancillary > chunks, such a gAMA and eXIf in this situation. Yes it does! Read the rules for PNG editors carefully. They don't say it is legal to write a PNG with the wrong metadata, even if you include the right metadata as well and it would be contrary to common sense to suggest that that was the intention. We had to made an explicit exception to common sense to allow eXIf to introduce incorrect metadata and the result is intractable. A editor that conforms to those rules with eXIF (unsafe) *must* either update the relevant tags in EXIF, delete the relevant tags or delete the whole chunk. There is no other possible interpretation; it's exactly the same as if an editor edits a PNG with an iCCP chunk. The only argument against unsafe-to-copy was Cosmin's assertion that the chunk has to be safe-to-copy because the textual metadata is too valuable. Everyone else who expressed an opinion effectively argued strongly against the safe-to-copy semantics. (Remember, that is why you put the line in the "security" section; the thumbnail argument.) I never regarded Cosmin's argument as compelling but none of the anti-thumbnailers argued against it. I accept Adam's reasons but I don't think adding more words that no one will take any notice of helps. The default has to be to delete the chunk; something that understands it can easily maintain it; exiv2 is a complete, adequate, library implementation and so is exiftool. John Bowler ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot