Cross posted to
nbajax@... and
dev@...I
wanted to get an idea of what the expected level of functionality the
javascript debugger is going to be capable of offering when debugging
web pages generated from templates.
I'm using Ruby on Rails and all responses returned to the browser
are constructed from erb files (although other templating systems are
also available). Basically a erb file is a normal HTML file with
support for embedded ruby code. Rails constructs the responses by
merging many erb files to create the resulting response to the user.
Often the same erb file is called many times with different parameters
for constructing things like lists etc..
Is it expected that developers will be able to place breakpoints in
these erb files in any location and have the debugger resolve the
locations of the many source files that may have contributed to the
resulting page the user is currently viewing? (I'm not sure that this
is even possible without injecting debugging specific markup into the
served page as the page construction is a lossy process like
compilation).
I have experimented with setting break points in a default Rails
application and the results are less than stellar. I can't get the
javascript debugger to reliably stop at a breakpoint in a erb file. I
can get it to work well for static HTML files if I use the full url
e.g. localhost:3000/index.html rather than localhost:3000. I can also
get breakpoints working with included JS files, but only if they don't
have parameters(isses #140442)
In the past I have got the debugger to stop at some breakpoints in
a rhtml file that was a single rhtml with no layouts or partials (not
the Rails default way of doing things). But the debugging features were
broken after it has stopped and there was nothing you could do except
cancel the debugging session. I haven't been able to repeat this
limited success with later builds and using erb files.
I'm not privy to how the javascript debugger works, but from the
little bits I have gleamed I think it does some sort of matching the
URL to the files on disk. I can't see how this is going to work with
Rails as this link is tenuous at best. Especially when using restful
routing (the default and preferred method) with nested resources. The
generated URL do not match the file system structure at all.
e.g.
localhost:3000/people/1/
addresses would call the AddressController and load the view /app/views/addresses/index.html.erb by default. Rails knows to pull apart the URL and supply the person ID of 1 as a parameter to the addresses controller.
Some guidance on what level of support/behaviour you guys are aiming at would be very helpful.
Cheers,
Anthony Richardson
