Re: JLink Problem, Fixed More or Less
- To: mathgroup at smc.vnet.net
- Subject: [mg87493] Re: JLink Problem, Fixed More or Less
- From: David Bailey <dave at Remove_Thisdbailey.co.uk>
- Date: Fri, 11 Apr 2008 05:57:03 -0400 (EDT)
- References: <ftkbdh$bi8$1@smc.vnet.net> <ftmttj$4mo$1@smc.vnet.net>
Kevin J. McCann wrote: > Interesting. Here we have a related problem without the locking up of > Mathematica. Every time a user uses F1 to get help a strange JLink error > message hpops up, often in a super large font that eventually reverts to its > usual. Happily I have ignored this without ill effect. It is annoying > though. > > Kevin > > Helen Read wrote: >> Several months ago I reported a problem with Jlink that has become >> rampant on University-owned and home computers since approximately >> December or January. Basically what happens is that some / most of the >> time when launching Mathematica, Mathematica will hang, and eventually >> pops up an error about failing to connect to Jlink. Usually you can >> ignore the message and go on, but sometimes worse things happen unless >> you restart Mathematica. We have also had problems with Mathematica >> hanging when trying to open up the Documentation Center (though I think >> the Documentation Center issue may be a bit better with the 6.0.2 release). >> >> We never saw the Jlink problem until we began switching machines from >> Norton Antivirus to NOD32 in December/January, at which time the problem >> became rampant, even on quite speedy new machines. We were hightly >> suspicious that the problem was related to NOD32, but WRI Tech Support >> never was able to tell us what the issue was or how to fix it, and >> several people from WRI assured us that NOD32 was not the culprit. >> >> Well finally, the whole thing had become so aggravating that we tried >> temporarily disabling NOD32 on a few test machines. What we found was >> that the problem stopped when NOD32 was disabled, and returned when >> NOD32 was turned back on. We then tried adding the Mathematica >> installation directory to the NOD32 exclusions, which seems to fix the >> problem. >> >> All's well that ends well I guess, and I suppose we should have trusted >> our own instincts about NOD32 and tested it ourselves far sooner than this. >> > I would install Mathematica on a clean Windows machine, and check that it is glitch free, and then add back the other software piece by piece - starting with the antivirus code. I use J/Link a lot, and I have never seen any of these startup glitches, so I guess WRI were in the same state of unawareness when they decided it was safe enough to use for essential tasks such as displaying help! David Bailey http://www.dbaileyconsultancy.co.uk