Re: [boost] Javascript?

From: William Kempf (williamkempf_at_[hidden])
Date: 2002-09-10 06:11:59

>From: "Paul Mensonides" <pmenso57_at_[hidden]>
>From: "William E. Kempf" <williamkempf_at_[hidden]>
>
> > That explains why you've got code to detect the browser in your 
>Javascript?
> > Sorry, but this stuff isn't really portable, and "portable" usage means
> > detecting the browser, which in and of itself is often done in a manner
> > that's not 100% fool proof, is likely to break with the next version of 
>the
> > browsers, and won't cover all target browsers in any event (since Boost 
>is
> > totally open on this).
>
>I have no major problem removing the Javascript from the PP libs.  It is no 
>big
>deal, but nevertheless, the pp-lib documentation is using compliant code 
>that
>works on the two most used browsers (by far) MS and Netscape.

Which versions?  Do you know how many older versions of both are actually in 
use?

I'm not trying to be flippant.  The thing is, I have to write HTML 
applications where the use of Javascript is basically essential, and I 
therefore have some experience with what it really takes to do that in a 
"portable" manner.  And in my case we have the option of requiring the use 
of a subset of browsers/versions.  We still get the maintenance complaints 
that the apps are broken and/or display poorly and have to spend time 
diagnosing that the user is running an unsupported browser/version.

With Boost, this is much worse, since we don't (can't/won't) specify what 
browsers/versions are supported, and our documentation should also "play 
nice" with non-browser utilities which are VERY likely to not support JS.

>  Also, it doesn't
>do any browser detection at all.  Things are getting better as far as 
>compliance
>to the DOM.  This, as with tabs in source files, is not really a big deal 
>to me,
>but I find it kind of ironic that Boost seems to abhor evolution is nearly
>everything but C++ itself.

I don't see that as true.  Boost is actually pushing the evolution of some 
things.  For instance, Boost.Build is pushing the evolution of software 
construction beyond the simple make.

>  I understand interests of portibility and all that,
>and I'm not saying it isn't valid.  However, I now have to dumb down the 
>utility
>of the interface for browsers that aren't compliant.  I though we cared 
>about
>standards here?

Yes, we care about standards, but we live in reality.  For instance, look at 
how much effort is put into making libraries compliant with compilers that 
don't meet the standards.  In this specific case, JS simply isn't portable 
enough, standardized or not, and since Boost's focus is on C++ library 
development and not HTML, it's more pragmatic to simply limit the use of web 
standards to the strictly portable subsection.  If we don't do that, we'll 
spend way WAY too much effort on fielding complaints and tweaking the 
HTML/Javascript to work with browser or utility X.

> > > 6. According to recent info in net ~95-97 present of browsers domain 
>is
> > > covered by ie and clones. Among the developers numbers are different 
>of
> > > course but ie still prevail I think. All others browsers try to keep 
>up
> > with
> > > de facto standard ie. Though I would not want to discuss this point 
>too
> > > much.
> >
> > I'm glad you said that, because usage patterns like this mean NOTHING to
> > this context.  If I were designing an online website for banking, for
> > instance, I might be swayed by such data, but if there's a single Boost 
>user
> > who can't make use of the documentation because of Javascript, well...
> > that's one user too many.
>
>Yes they do.  They may not be the most important thing, but presentation 
>*does*
>matter--even for Boost.  Also, that is like saying, "All I have is notepad, 
>and
>printed out HTML has a whole bunch of tags in it."  According to that 
>logical,
>why don't we just use plaintext?  I don't disagree wholly with the idea of
>banning it, but at the same time presentation and ease of use do matter.

I never claimed presentation didn't matter.  I claimed usage statistics 
didn't matter.  Different argument totally.  Even if 99.999% of every user 
was running IE on Windows XP, that would still be irrelevant, since there's 
still users using something different.  Boost documentation has to be 
maximally portable, not just usable by the majority.

And don't forget that there's more to this argument than just presentation 
in a browser.  Users will run the HTML documents through all kinds of 
utility programs as well.

Bill Kempf
williamkempf_at_[hidden]

_________________________________________________________________
Join the world’s largest e-mail service with MSN Hotmail. 
http://www.hotmail.com