Group Abstract Group Abstract

Message Boards Message Boards

4
|
39.7K Views
|
34 Replies
|
26 Total Likes
View groups...
Share
Share this post:
GROUPS:

Stability of Mathematica 10

Posted 11 years ago

I have been a regular and enthusiastic user of Mathematica since version 3. I am very concerned about the stability of Mathematica 10.0.1. It crashes when attempting spell check. It often hangs and then crashes when using == (access to Wolfram Alpha). It crashes when editing complication 2D expressions, especially of delete is used. Perhaps this is related to the undo feature. Is my installation somehow damaged, or are other users experiencing this type of difficulty? My frustration level has risen to the point that I am seriously considering switching to a competitors software. Mathematica's capabilities and feature set are tremendous, but that is not valuable to me if it crashes frequently. I would really like to know what other users are experiencing.

POSTED BY: John McGee
34 Replies

I had a similar flakey problem and ended up communicating with Mathematica Support who suggested the following:

  1. With Mathematica open, select 'Preferences...' from the 'Edit' menu. Then select the 'Evaluation' tab in the window that pops up.  Try hitting the 'Reset to Defaults' button and see if the computation will work then. 

  2. If the above does not resolve the issue, try resetting Mathematica to its default settings.  For instructions on doing so, please see     http://support.wolfram.com/kb/3274   The first item did nothing to help at the time BUT the second one not only fixed my problem then but recently also another instability problem with using the Graphics -> Drawing Tools. - That's 2 for 2 in correcting instability problems. Only problem is you have to erase your preferences file (save first !) and re-register your product -not a big problem and it is soooo nice to get rid of that instability or buggy behavior.

David, I read your wish and while I was redoing my keyboard shortcut code I added a "Command Option M" keyboard shortcut ("Cmd Alt M" on Windows, "Mod1 Mod2 M" on Linux) here. It is a bit slow (on MacOSX, faster on Windows), but it works. I am sure the good folks at Wolfram will implement something like this for some future release. It is not that hard to do.

POSTED BY: Rolf Mertig

A 10.1 would be great if the company undertook to fix ALL outstanding bugs and introduce no new features.

POSTED BY: Mike Honeychurch

Perhaps this is a case of too many projects underway at once. I saw Wolfram people using 10.0.2 for their talks at the Wolfram Technology conference in October, so it is not 'vaporware', and there are real fixes to problems. I would have thought that 10.0.2 would have been out already.

Workbench 3 won't be out until 2015 (early), from what I have been told, but I have no information about a version 10 compliant CDF player. David Park, in his discussions on this thread, has convinced me to give workbench another shot, but I cannot share any of my version 10 work with non-Mathematica owners until there is a CDF player that they can use.

From the viewpoint of a user, I think it is time for a semi-major version -- say 10.1.0 -- that would consist of nothing but bug fixes and bringing CDF player, Workbench, etc. into sync with the main product.

I would include in this 'bug fix' release such things as the Cocoa front end for OS X that is way overdue. I am fully cognizant of the technical difficulties, but the Carbon APIs that are in the current front end have been deprecated for at least 5 years. I am sure that there are similar issues for Windows and Linux. This discussion started on the topic of stability, and from all anecdotal information, the OS X version is far more stable than Windows or Linux.

We have not had a '.1' release since version 5.1; they were fairly common before then. While naming versions is somewhat arbitrary, a 0.0.1 update indicates that there are minor changes, or stuff that did not make the cutoff. A '.1' release indicates, to me at least, that significant issues with the product have been rectified without making substantial changes in the program's 'footprint'.

I think that this is exactly what Mathematica needs.

Well, we have been promised a desktop version 10.0.2 "soon", since October or so. So it will come, well, soon. Like PlayerPro 10, also promised soon, this year, which proves that they are indeed working to fix bugs and have nice new features.

What do you mean by "released"? It still says "BETA" on the programming.wolframcloud.com site. Some things already work very nicely, but, e.g., ListPicker, or uploading an Excel file does not. Also, while JLink works, MathLink and LibraryLink do not work on the cloud, which is unfortunate, but understandable for security reasons. Of course one would want to fix this in a private cloud version. But that version seems to be not available yet, hopefully soon.

Multiple undo is nice (though has been there for a long time for proper .m files in Workbench or IntelliJ or vi), but if this is the cause of crashes I will switch it off immediately once I use more Mathematica 10, which is supposed to be released soon (10.0.2, I mean).

POSTED BY: Rolf Mertig
POSTED BY: Ulrich Mutze
Posted 11 years ago

Again, I envision that the best form of applications would be like advanced textbooks but with much greater power and flexibility.

That sounds reasonable but I have not got the idea. Could you be more specific about how such an application may be organized and worked with?

POSTED BY: Alexey Popkov
Posted 11 years ago

It would be great if Wolfram Research could provide a means for users to provide Mathematica projects that are more involved than Demonstrations.

I agree absolutely. We all need some basic functional we can rely upon. Currently we have some extremely limited reliable basic functional which seemingly was not extended since the Mathematica version 3.

POSTED BY: Alexey Popkov

This may be a case where the squeaky wheel gets the oil.

Based on my discussions with the Wolfram engineers, they are aware of problems -- certainly my issues. They do not always get around to fixing them as fast as I like, but multiple contacts could not hurt.

At various times, I have gotten a 'boilerplate' type response from technical support. If this type of response does not meet your needs, you should respond with more information or clarification. I can tell that this strategy will often get a problem before more people, and that has to be a good thing.

It is my belief that a major part of WRI's resources have been allocated to Mathematica on-line and related projects. Now that that project has been launched, I am hopeful that some issues with desktop Mathematica (which should also affect the on-line version) will be addressed.

Perhaps more than most other programs, Mathematica will thrive only if enough people are passionate about the program and its potential. Having been with Mathematica since version 1, I can tell you that the company does care, and eventually, major issues do get addressed (multiple undo, anyone??).

Thanks for the information. I was not aware of some of these options.

Once a Mathematica 10 compliant CDF player has been released, I will check them out, using my wife as a beta tester.

Posted 11 years ago

I am a very young Mathematica user, started with Mathematica 9, using it under Windows 7. I don't have stability issues unless Mathematica tries to connect to the internet and fails because the proxy server is not set properly. Then it seems it is dealing with a wait / timeout behavior that finally forces me to kill Mathematica with the task manager. Beside that I did not experience stability issues.

However, I already reported multiple issues to the Mathematica support. Issues which could be considered minor in general but are critical to me. I have the impression that wolfram does not take these inputs seriously. When contacting the support they usually provide first feedback that the report was received but after that things go silent. Of course no solution is provided eventually.

For a software that costs multiple thousands of € for commercial user that is quite an issue on its own.

POSTED BY: Mathias Breuss
Posted 11 years ago
POSTED BY: Alexey Popkov

Have not noticed stability problems in V10 on Yosemite and prior to that Mavericks but admittedly did not use it much due to the amount of bugs present and the lack of Player Pro. I am still on V9 95% of the time but want to shift 100% to V10 when these issues are fixed because I like some of the new features. It remains a source of frustration that bugs go unfixed for multiple versions. In the V10 installation I have tried (10.0.1) there are bugs reported in V7, V8 and V9 still present.

POSTED BY: Mike Honeychurch
Posted 11 years ago

I wonder what users think about having multiple Menus on multiple notebooks?

What I really like about the new-in-v.10 multiple windows interface is that it provides a workaround for the long-standing bug of empty list of opened Notebooks in the "Window" FrontEnd menu: now the individual Notebooks are also available from the Taskbar and I no longer have to rely on the FrontEnd menu.

Just presenting a "palette" of choices (such as PlotThemes) to users is not sufficient. It vastly underestimates the true power of Mathematica with its active calculation and dynamic controls.

I agree absolutely. The PlotThemes-style of development feels like step backward in the design of the system: it does not expand the possibilities but provides a limited number of hard-coded choices which are already simply outdated in their design (and always will for the obvious reasons!). It is not an expedient direction of development. What I expect in the future versions are new ways to construct graphics from other graphics. In particular I lack vector CropByMask command which would allow to crop a vector object by some vector mask defined, for example, as a Polygone (keeping everything in vectors of course!). I also lack the working FullGraphics function (which starting from version 6 simply does not work in the most of cases). I also feel that the Ticks functionality should be completely redesigned to allow much more flexibility: easy control on the distance from tick marks and the axes/frame, built-in functions for generating Ticks specification with at least that functionality which currently provides the CustomTicks` third-party package (what a shame that for such a basic task we still have to install third-party packages!).

POSTED BY: Alexey Popkov

I have raised the issue of access to lower-level routines with Wolfram engineers at various times. With Xcode (and I am sure other IDEs), one always has access to the file headers to get detailed information about functionality, and I think that Mathematica or Wolfram Language needs the equivalent. The Wolfram engineers have access to them, because they use them in their own presentations and when answering technical questions. Given that the users of these primitives are likely to be well beyond the novice stage, the level of documentation can be minimal.

Perhaps there could be a developer program that would provide additional documentation and access. I remember that the Wolfram Technology conference started out as the Wolfram Developer's conference. The company was a lot smaller then -- and the Mathematica coder base was much smaller -- and the conference gave you access. I think that this idea can be built upon and formalized.

Even having all possible options show up in the code completion popups (e.g., options for "Appearance") would be useful. A preference could be set to indicate the level of detail the user wanted, so the list of choices would not overwhelm a neophyte.

Mathematica 10 appears to have caused more problems for Windows than for Mac. I considered it bad enough on Windows 7 that I reverted back to Mathematica 9.

I wonder what users think about having multiple Menus on multiple notebooks? It may be connected with the new Undo and at first seems like a good idea in itself. But it has one major problem: it's no longer possible to minimize Mathematica with one click as in previous versions. One might want to do this to temporarily put Mathematica aside and make screen room to work on some other task.

There is another problem with this concerning the use of side windows. Using a side window is useful if you want to present a display that remains in view while scrolling the notebook. There are two principal types of side windows, notebooks and palettes. If you want to minimize Mathematica you now have to minimize all the notebooks and then the palettes are automatically hidden. You can put dynamics and graphical displays in palette windows - except that you can no longer rotate 3D graphics with the mouse. I did submit a bug report and notebook and it was sent to the developers. I hope they fix it in 10.0.2.

Another concern that I have is stability over Mathematica versions. Beyond the use of Mathematica as a super graphical calculator scratchpad, I believe that a natural usage would be the writing of literate notebooks and applications consisting of literate notebooks, packages and documentation. Mathematica has tremendous advantages in this area but I don't know if it's very much exploited. These take much more work than just scratchpad calculations and a lack of long term stability may mitigate against it.

One trend in Mathematica that I find disturbing is the introduction of more "set-piece" high level routines without easy usage or even access to lower level routines. (One example might be graph theory, which combines algebra with graphics. There are many other rough patches, many of them that revolve around graphics.) It is all right to have "set-piece" routines for common things that users do. But they should be built on lower level accessible routines.

Just presenting a "palette" of choices (such as PlotThemes) to users is not sufficient. It vastly underestimates the true power of Mathematica with its active calculation and dynamic controls. You can't just present users with a palette of animated (say) 19th and 20th century graphical presentations. The universe of concepts or data that one might want to present is too large and the possible methods of presentation too great. We have no idea what great new techniques might be invented. Users need easy methods to build these up using primitives. It should be more like writing on a sheet of paper.

Posted 11 years ago

I'm on a Win 8.1 PC. I've submitted two problem reports to WRI (both confirmed). One would result in a front end lock up. The other would result in a PC lock up. I haven't experience your problems with crashes, but I have experienced timeouts when using WolframAlpha. I've read about other issues like the spell check problem. I think this release was not up to the usual WRI standards. I know WRI is working to correct the issues.

POSTED BY: David G
Posted 11 years ago

I've had this problem too. If it's caused by Undo (I've thought of this also), it would help if there were a command to clear the Undo cache, that could be used occasionally during a process (such as, in my case, a long sequence of non-local image transformations) you're afraid might cause a crash.

(Note that occasionally saving the notebook doesn't help much, because it's often during a save that the crash occurs.)

POSTED BY: Collin Merenoff
Posted 11 years ago

I don't think the details of his platform would be nearly as helpful to anyone attempting to reproduce this or possibly even diagnose this as a notebook containing n independent specific examples with directions to 1: reboot the machine, 2: boot Mathematica, 3: open this notebook, 4: evaluate a single specific cell in this notebook and 5: based on his repeated testing and for for his platform evaluating this cell will crash Mathematica about x% of the time. Since he has, more or less, reproducible crashes then this should be straightforward to do, it is the ones that can't be reproduced or even narrowed down to a specific example that are the harder ones. Otherwise this sounds like "I typed some stuff in, some stuff came out, some stuff is wrong, what?" that have been posted by some users in the past.

POSTED BY: Bill Simpson

John,

The spell check crash in Mathematica 10.0.1 has been reported to support@wolfram.com (and it has been fixed in the upcoming Mathematica 10.0.2 release).

POSTED BY: Arnoud Buzing

John,

I am running M10 under the latest Mac OS X and have no such stability issues. As others have suggested, more details on your platform would help.

David Mackay

POSTED BY: David Mackay

I am using windows 8 and it is fine. Looks like something is really wrong with higher version of windows. Ulrich Mutze, I think you should contact Wolfram help center.

POSTED BY: suvadip mandal

I'm using Windows 8.1 and my experience is not too different from John McGee's.

POSTED BY: Ulrich Mutze

Actually the note was meant for John McGee ;-) . And, yes, the crash reports are kind of like looking at the insides of sausage and trying to figure out what animals contributed....

POSTED BY: David Reiss

Of course -- including the crash report generated by OS X. I've only had 2 or 3, and they aren't strictly deterministic. I don't think that they are random either. I'm not good at reading the goat entrails from the crash logs. ;-)

It has been stable for me as well (OSX).

But, if you are able to create any reproducible crashes I hope that you reported them as bugs to WRI.

POSTED BY: David Reiss
POSTED BY: John McGee

Perhaps you could tell what OS you are using. I am using OS X (10.10), and see none of your symptoms. I have had a couple hard crashes (sent info to Wolfram Support), but re-opening the notebook was successful.

From a recent virtual event, someone at Wolfram indicated that there will be a 10.0.2 update "soon". Perhaps this update will resolve your problems.

Hi, I am a Mathematica user too and I use Mathematica 10 but I do not see any discrepancies regarding crashing of the software. It is doing well. Perhaps your installation was not done properly or some files were missing.

POSTED BY: suvadip mandal
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard