[Glastonbury] Java

[Andrew M.A. Cater]

On Thu, Oct 30, 2003 at 06:05:00AM -0000, Sean Miller wrote:
> 
> Erm, it is each browser that gives the "Click OK to download" message...
> presumably *it* would know in what directory it has just looked to ascertain
> the plug-in is not present? Or is that over-simplistic??! [I think *not*]
> 
This is over-simplistic. Don't anthropomorphise the computer. The 
browser is merely rendering the page.  The plug in may have been 
assumed to use one fixed path e.g. ~/.netscape/ or /opt rather 
than /usr/local and the install routine may have been hard coded 
accordingly, for example.

> > Whose JRE and which version? 1.1/1.2/1.3/1.4?? Sun/IBM/Microsoft/Blackdown
> > Linux port/GNU Classpath stuff??
> 
> Erm, again... when you click "OK" it downloads whatever it is it thinks it
> needs... I believe Sun is most likely, but yet again it is the browser that
> makes the choice therefore it is not unreasonable to expect the browser to,
> having made that choice, complete the install without the user having to
> mess around copying files manually afterwards.
> 

See above.

> > All of the above may be relevant.

Permit me a digression which may help clarify my point.

Someone comes to me and says "the exhaust on my car is loud and I
think it's making a different noise"

My response may depend on the information I've been given: it may also
depend on who I'm talking to.  Others may also have other good advice.
All of the following are potentially valid points, for example.

"It might be a good idea to shut the car window again" [For Sean Miller 
:) ]

"It might be good to take it to Kwik-Fit and get them to have a look at 
it" [For my mother]

"Take a quick look underneath.  If it's a small hole that's blowing, you
may be able to patch it with an exhaust bandage and something like 
Holt's Gun Gum. If it's on the silencer/baffle, you may need to take it 
to a garage and have a section replaced.  If you can smell strong 
exhaust fumes within the car or performance is poor then the hole may 
be at the exhaust manifold on the engine block. You'll need a garage to 
look at that" [My father and late uncle's responses]

"Find me some pipe and I'll make you a new one if I can't braze on some 
metal over the hole" [The Scrapheap Challenge teams / more or less any 
mechanic in rural Malaysia/Indonesia]

Linux is very much in this paradigm.  You get conflicting advice and 
expertise - if you have more experience / know who to ask and how to 
ask, you can narrow down the problem more quickly and educate yourself 
into the bargain.  

I was trying to ask relevant questions in an effort to narrow down 
the scope of the problem.  Red Hat 9.0, for example, has a problem with 
the Sun JRE -it's not particularly Java related per se but rather 
because the NPTL threading implementation which Red Hat have 
incorporated into their 2.4 kernel and backported from kernel version 2.6 
doesn't work with this Sun application because the Sun application is 
not quite compliant.

> 
> A trait that I have noticed with both yourself and Martin is that if
> somebody makes a perfectly reasonable comment regarding the usability of
> Linux programmes versus Micro$oft your first reactions always seem to be to
> bombard them with versions, applications and various strange techno-babble.
> Of course this does nothing to actually answer the original point which was
> that in Micro$oft browsers if it says "click here to download plug-in" you
> click there and five minutes later you have the plug-in.... in Linux, imho,
> you click there, it downloads, you close your browser, you return to the
> page and it says "click here to download plug-in" again....
> 

First of all - you may have missed the point.  It appears to have been a 
permissions problem as I suggested in the first place. The 
effectiveness of "download this plug in and it will install" varies, 
even amongst MS operating systems.  It may depend on a whole host of 
factors like which service pack you have installed, for example. Java 
is not a particularly good example in any event.  Java has been under 
active development: API's have changed such that you may need to rebuild 
your code for it to work. "Write once, run anywhere" has been hamstrung 
by various subtly different interpretations and implementations of the 
Java environments.

> 
> As I said, I would expect Mozilla to know where it is looking and download
> to the appropriate place. That does not seem, to me at least, to be
> particularly unreasonable.
> 

Most plug in programmers may assume one or two browsers and one or two 
"standard" locations.  The "standard" browser may change over time: I 
was trying to make the point that it is very hard to produce canonical
locations for all browsers and it is probably unreasonable to expect any
arbitrary browser to "just work" at all times.

> 
> Also, from memory, I think that the "click OK to download" functionality
> *does* actually work on Windows Mozilla - I think it is only the Linux
> version that has this strange "trait" so, perhaps, my original concern re.
> competing is actually going back to a Linux desktop/Windows issue again...
> is it the Linux developer community that are to blame, perhaps? If so, can
> we borrow some Windows developers to teach us how to make things
> user-friendly? Who knows... perhaps once they've worked in it for a while
> they won't want to go back?? ;-)
> 

Please re-read what you have written above.  It's not necessarily 
simply a "Linux/Windows user friendliness" thing - it is sometimes a case of 
potentially mistaken apprehension / over hasty responses to the presumed 
causes (or simply something that punches peoples buttons and causes them to 
give stereotypical responses :) )

["We shall make you read, so that you shall not forget" Surah 87, the 
Holy Qur'an]

Apologies for a long-ish post :)

Andy