Jim Land (NO SPAM) wrote:
"VK" <sc**********@yahoo.comwrote in news:1163278275.629360.236530
@h54g2000cwb.googlegroups.com:
Peter Michaux wrote:
A subset of SQL grammer is what I would like to implement in
JavaScript.
<http://trimpath.com/project/wiki/TrimQuery>
Great zot! Trimpath extends Javascript with SQL queries! It's a
lightweight GPL open-source component that gives you the power of SQL
queries while running Javascript in a web browser. Written in Javascript.
I was just as excited when I saw it the first time. A DBMS in only 600
lines of JavaScript. I think it is too good to be true and some
investigation into the code has shown it too have problems. The idea of
a DBMS certainly could happen in a browser but is it a good idea?
A DBMS in the browser is appealing but I am conflicted about the
appropriateness of such a beast. How much thinking should a browser be
able to do? It seems like a slippery slope where the desire for a
little bit of cool DHTML could lead to way too much browser brains. For
example, in a DHTML data grid if the user can filter which rows are
visible or in which order they appear then custom code has to slowly be
written to manipulate the data. When the number of user options for
data filtering and sorting increases enough it would be better just to
write something like TrimQuery. With the innocent wish to sort a list
now download times have increased with more data and libraries to be
cached in the browser.
It looks like Gmail leaves all sorting, filtering and pagination up to
the server since the server has all the tools needed like a DBMS to
program this quickly. This also keeps individual page load times low. I
would not make sense to load 20 000 emails into my browser just so I
can search through them quickly. However when there are only 30 items
of data and only a few sort options then people want to delegate the
responsibility to the browser. Where is the line? It seems
fundamentally wrong to give the browser so much power but I don't know
how to articulate clearly when the line has been crossed. Perhaps it
is in the grey area.
There is also the problem of distrubuted data that needs to be
syncronized with the server data and what conflicts that might occur.
I think Gmail got this right.
This is primarily why I started the other thread called "The future of
applications in JavaScript?" I don't know if moving in the direction of
TrimQuery-type applications is a good idea or not. If it is a good idea
than we need more traditional software development tools like a parser
generator for developing JavaScript apps.
Peter
