in message to "Davey" concerning
Opera Bug #CORE-22089
Link?
I can't picture it.
I don't know. Post an example.
NP.
Thank you for your interest in this post.
No, thank you.
Yes. Says the problem is present up to 10.something, which I assume
means everything prior.
this is my project which suffers from this bug,
http://www.daveyerwin.comIhave commented out the
workaround which relieves this problem (no verticl scrollbar).
To observe the problem adjust the vertical size of the Opera
browser.
And do what? Didn't do anything special in 9.27. Vertical scroll bar
appeared as expected.
I don't know what you mean by "offending DIV" or why you would want to
infer from Opera that scroll bars are missing. If you want to know
the overflow style of the HTML element, check the computed overflow
style of the HTML element.
Any time you think to use browser sniffing, even for the smallest of
decisions, stop. Even if it wasn't ten years out of date, it is
clearly an unreliable, time-consuming and invariably doomed strategy.
But like Charlie Brown trying to kick the football, Web developers
have been trying to make it work all decade.
MS has traditionally
been their biggest Lucy (despite the snail's pace production on IE),
but those were salad days compared to what is happening now.
Assuming you used browser sniffing for this, if and when Opera fixes
the problem, you could write off users of all but the latest Opera (a
popular strategy). Declaring that your script might work in older
versions, but no time will be wasted (by you) testing them is another
popular spin (passing the time-wasting on to the end user). I guess
if the testers can't see regressions, they don't exist.
Alternatively, you could add another fork for the Opera version number
(and still miss other browsers that have this quirk). Then what if
they break it again? Browser sniffing code tends thrash about as
observations are logged. Read the comments and change logs for such
code and it is apparent that decisions are based on whatever the
"offending" browser was noted doing at one specific moment in time.
It should be obvious that each such "solution" is the coding
equivalent of a band-aid. Usually they just cover up an underlying
problem (or confusion) in the design. Patch designs like this long
enough and they become virtually unrecognizable (that's usually where
I come in and the first thing I do is tear off all of the patches so I
can _see_ the problems).
Even on the tiniest projects, the logic eventually becomes so complex
and unwieldy that nobody will touch it for fear of disturbing the
delicate "cross-browser" balance. But the browsers keep changing, so
inevitably the scripts become unusable to all but nostalgia buffs.
