[Openicc] colord information

Chris Murphy lists at colorremedies.com
Fri Feb 11 14:51:20 PST 2011


On Feb 11, 2011, at 4:33 AM, Richard Hughes wrote:

> On 11 February 2011 10:34, edmund ronald <edmundronald at gmail.com> wrote:
>> I like virtual users for interface design work, because they focus (or
>> detour) the imagination in a way just staring at the screen cannot
>> achieve.
> 
> The problem with virtual users is that it's very easy to devise
> oddball use cases that doesn't represent the majority, or that you can
> easily test with. This leads to feature creep.
> 
> I think feature creep is a *huge* problem in open source software.

Yes like proofing. I say drop it for now. Both soft and hard proofing. Let's get display and printing working well. There are very few instances where users need to be polled on settings or preferences.

To me a huge elephant in the room is Hal's explanation of how complicated it is to get all the different libraries responsible for file format support, to honor embedded metadata, and then the application honoring/preserving it, and then making sure when it asks various libraries to re-write out the file that the metadata is preserved. It's a huge problem, it took years to fix it on Mac OS. It's still not entirely fixed, we still have "safe" applications and "unsafe" applications that will strip metadata. And it's really annoying to me that this isn't something engineers have looked at solving from a much bigger picture point of view, just because they don't translate dropping certain pieces of metadata with data loss. It's in effect a corrupted file. We'd never tolerate this sort of behavior from a file system. it's one of the things that screwed PNG's color management implementation.

And on open source, it may be next to impossible to get all the libraries that deal with image file types to not drop metadata. But I seriously would be using vitriolic, caustic language toward any application or library that is not preserving data as a matter of course. These are nothing short of saboteur applications. I'd ask them to be stripped out of distributions for default installation, and ask users and other developers to stop using those tools until they get fixed, because they are fundamentally broken.


Chris Murphy


More information about the openicc mailing list