[rkward-tracker] [ rkward-Bugs-3056567 ] NEEDINFO: Crashes/freezes on xmlTreeParse

[SourceForge.net]

Bugs item #3056567, was opened at 2010-08-31 09:08
Message generated for change (Comment added) made by tfry
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=459007&aid=3056567&group_id=50231

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: data handling
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Nobody/Anonymous (nobody)
>Summary: NEEDINFO: Crashes/freezes on xmlTreeParse

Initial Comment:
This works fine in a stand alone R console:
   library(XML)
   xmlDoc <- xmlTreeParse("http://www.unimod.org/xml/unimod_tables.xml")

rkward (fresh checkout/build from svn) crashes or freezes on it (couldn't figure out why one or the other is happening).

----------------------------------------------------------------------

>Comment By: Thomas Friedrichsmeier (tfry)
Date: 2010-09-01 16:57

Message:
Thanks for reporting this issue.

I can reproduce the freeze (see below), but not the crash. Could you give
some more detail on this?
- How much of the time does it crash?
- Does it crash without further interaction, or only after trying
something else (e.g. quitting RKWard)?
- Do you get a "KCrash handler" window? If so, could you please attach the
backtrace from that?

Regarding the freeze: It's actually not a freeze, but it's just slooooow.
Basically the problem is that xmlDoc contains a lot of nodes, and
extracting each single node is very slow for some reason. This is only
partially the fault of RKWard: The slowness probably comes from the
XML-package code. The problem in RKWard is that it is too proactive about
gathering information about xmlDoc, and therefore triggers all those slow
operations. Roughly speaking, what RKWard does is the equivalent of
   str (xmlDoc, list.len=10000)
which takes a very long time in a plain R session, too (note that the
bottleneck is not the printing of all that output, but gathering it in the
first place).

I plan on making RKWard less pro-active in gathering this information
about objects, but this is not going to be trivial, and will not happen too
soon.

For the time being, the generic advice to work around this type of problem
is to use local variables (e.g. by placing your code inside a function,
instead of at the top level).

Regards
Thomas

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=459007&aid=3056567&group_id=50231