You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Should we add the possibility of adding an src attribute to the <source> element?
It can be useful as a shortcut if only art-direction is needed, or if CH is used to do res switching. (with the RW hint).
Alternatively, we can ignore it, since a single URL with no descriptors is perfectly legal with srcset.
I'm making this argument, but I don't actually believe it: I could see not having src causing author confusion, since it's a pattern they're used to using with video/source.
I like the idea of having src available (though we should keep no-descriptor srcset in play), but if it stands to complicate implementation any I could certainly live without it. Like @nwtn said: it will definitely be expected to work.
I'm actually torn on this one. On the one hand, I agree that devs will expect it to be there. On the other hand, why didn't we define srcset to be src? (Mainly because devs would expect src to behave like src)
Having 2 attributes won't complicate implementation much, but will uglify the spec & algorithms. Beyond dev expectations, there's no real need for 2 distinct attributes, so I don't know what's the best approach.
Since a valid src value is also a valid srcset value, one could make the
argument that src is only needed for legacy purposes. Since is anew element altogether it doesn't need that kind of legacy support. src isthen redundant no? Plus, it would dive home a little more that within a is not the same as within or
I'm actually torn on this one. On the one hand, I agree that devs will
expect it to be there. On the other hand, why didn't we define srcset to
be src? (Mainly because devs would expect src to behave like src)
Having 2 attributes won't complicate implementation much, but will uglify
the spec & algorithms. Beyond dev expectations, there's no real need for 2
distinct attributes, so I don't know what's the best approach.
—
Reply to this email directly or view it on GitHubhttps://github.com//issues/78#issuecomment-31490283
.
Since no one seems to have a very strong opinion that src must be supported, we're probably better off leaving it out for now. Closing, but feel free to re-open if necessary.
Activity
nwtn commentedon Jan 2, 2014
I like the idea of having
src.I'm making this argument, but I don't actually believe it: I could see not having
srccausing author confusion, since it's a pattern they're used to using withvideo/source.Wilto commentedon Jan 2, 2014
I like the idea of having
srcavailable (though we should keep no-descriptorsrcsetin play), but if it stands to complicate implementation any I could certainly live without it. Like @nwtn said: it will definitely be expected to work.yoavweiss commentedon Jan 2, 2014
I'm actually torn on this one. On the one hand, I agree that devs will expect it to be there. On the other hand, why didn't we define
srcsetto besrc? (Mainly because devs would expectsrcto behave likesrc)Having 2 attributes won't complicate implementation much, but will uglify the spec & algorithms. Beyond dev expectations, there's no real need for 2 distinct attributes, so I don't know what's the best approach.
uniqname commentedon Jan 2, 2014
Since a valid src value is also a valid srcset value, one could make the is anew element altogether it doesn't need that kind of legacy support. src isthen redundant no? Plus, it would dive home a little more that within a is not the same as within or
argument that src is only needed for legacy purposes. Since
On Thu, Jan 2, 2014 at 3:24 PM, Yoav Weiss notifications@github.com wrote:
Cory Brown
yoavweiss commentedon Jan 2, 2014
Since no one seems to have a very strong opinion that
srcmust be supported, we're probably better off leaving it out for now. Closing, but feel free to re-open if necessary.