tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Peter V. Gadjokov" <>
Subject RE: Case Sensitivity in URLs (was RE: BugRat Report #92 was close d(apparently by: Craig R.)
Date Tue, 12 Sep 2000 02:57:59 GMT
>Your USENIX quote is fantastic!

Glad you enjoyed it :)

>Case-insensitivity in a filesystem is a HACK. Will always be a 
>HACK and only compounds the useablility problems it tries to solve.
>It is also a "different design decision" too. So it looks nice on a
>bulleted marketing prospectus.

None of these (you think it is a HACK, you think it compounds usability
problems it tries to solve and you think it looks good on a bulleted
marketing prospectus) are good, justified technical reasons - in the
way you've presented them now, they are simply your opinions. And 
against the weight of your opinions, I have the _fact_ that the two
most widely deployed desktop environments are case-insensitive and,
therefore, most users are familiar and accustomed to such behaviour. 
I could also say that using the user-visible filename string as a
unique file identifier is a HACK (if you've read the rest of the 
paper, you'll see an outline of the HFS/HFSPlus solition - stable
file id's for programmatic use) but since many many platforms rely
on that assumption (Java included), you don't see me suggesting we
don't support it.  

>Many times in the design of generic frameworks and libraries,
>descisions have to be made to accommodate architectures that 
>tend to "deviate from the norm". Upon closer examination, turns
>out they deviate because the are crappy and ill-conceived. There
>are many such examples through the history of computing. This is
>one of them.

This may be true if 'the norm' is undeniably the optimal,
all-encompassing solution. In this case the norm is largely a matter 
of habit and acceptance of a widely deployed approach. There is 
absolutely nothing wrong with that, I'm simply suggesting Tomcat
should support _other_ widely deployed, common approaches. 

As to whether it's 'crappy and ill-conceived', again, this is an
opinion of yours I'll happily discuss and even be convinced of
if presented in a manner other than 'statement of undisputed 
truth'. I doubt the tomcat-dev is the optimal forum for that, 
as I doubt that the featureset of the tomcat implementation should
be driven by the aesthetic preference of the implementors at the
expense of supporting typical deployment environments. Again, such
a feature may be indeed inappropriate and I'll shut up about it 
the second I hear a well-substantiated reason - 'it's a HACK' is
not one of them. 


View raw message