Message Boards Message Boards


Natural language computation symbol appears unintentionally in V12.1.1

Posted 1 year ago
18 Replies
11 Total Likes

Just upgraded to 12.1.1 and find it impossible to enter the most basic expression or do any standard form action because some new Alpha thing takes over and destroys my code before I can finish it. Any ideas how to disable the new, unwanted feature?

I can revaluate any cell I have coded in the past 20 years BUT I cannot write new expressions using assignment or delayed assignment. The moment I enter a "=", Mathematica 12.1.1 turns it into some alpha thingy and I cannot do anything more with that cell. Even covering it back to Input results in an error message. example of what happens when I try to assign a value to chi

18 Replies

Hi David Barnes, I'm eager to test out the bug fixes and feature enhancements of Mathematica 12.1.1 but noticed that default install is in the Mathematica 12.1 directory so I cancelled the install. Did you find that the update replaced Mathematica 12.1? Mathematica 12.1 is much more stable for the functions I use and faster than Mathematica 12.0.

In Mathematica 12.1, a ctrl= is needed to use the WolframAlpha natural language feature. What does ctrl= do in Mathematica 12.1.1?

Posted 1 year ago

So I guess I assumed Wolfram was depreciating the desktop and trying to go mobile first, as so many other companies have done. What I found after taking a deep breath is that I can create a new file in 12.1.1 and enter a "=" assignment as usual and "[Ctl]=" Euler evaluates the normal entry for Leonard E. However, that is not what happens to any of my saved (100's) of nb files. Entering "=" converts the input cell into some alpha thingy that cannot be changed or evaluated. Entering "[Ctl]=" Euler returns a not-parsed error in some files and the normal output in other files.

I can create new input in the new nb file created under 12.1.1 then paste the expression(s) into an older file (say from Monday) and evaluate it. I can also write code in a text editor and paste it into a nb, then evaluate it correctly, so Mathematical is not brain dead, just unable to accept a typed assignment.

A work-around, such as "(2+w)/.w->3" generates another good result I've not seen before: result of entering

This was Mathematica install thirty-something for me, so I did it the same way I that has worked for decades. Thinking of uninstalling Mathematica with CleanMyMac and downloading 12.1.1 again. If that doesn't work, I'll try to find a file fixer utility or something...

Hello David, there is an unintended behavior in Mathematica 12.1.1 in which if a text cell contains a word not in Mathematica's dictionary (so it has a red underline) certain characters that trigger behavior at the start of a cell (i.e. = opening the Natural Language parser, > opening the python interface) will trigger that behavior when entered in the middle of an input cell.

The immediate workaround is to disable the spell-check in Mathematica's preferences under 

>Interface > Check spelling as you type

We hope to have this behavior fixed soon, but this should serve as a workaround in the meantime.

Posted 1 year ago

Good to know, but the problem occurs in Input cells, not Text cells. My solution has been uninstalling my second attempt to get 12.1.1 working and reinstall 12.1.0 The older version works like it did earlier this week, so I will stay on 12.1.0 until there is a patch.

To clarify, the presence of certain words in text cells causes input cells in the same notebook to exhibit this behavior.

Are you going to publish a new build of 12.1.1 once this is fixed?

Yes, we will.

A new 12.1.1 or a 12.1.2? If it's called 12.1.1, I bet many institutions will not upgrade their machines, leaving users frustrated and not understanding what happened.

There is a new build of Mathematica (M-OSX-L-12.1.1-6958981 for macOS) available in the User Portal that fixes this issue. I Imagine that there are new builds for Windows and Linux, but I have no way to tell.

It appeared on 6/23, so If you DLed earlier, the new one is the one to get.

I tested it, and the issue has been resolved (with the usual caveats -- only checked macOS)

Unfortunately it is also called "12.1.1". Incrementing the version number would have been much harder due to various considerations, simply replacing the builds in the user portal is quite simple for us in comparison.

Indeed, the builds for all three platforms (macOS, Windows and Linux) were replaced on 6/23 with ones that have a fix for this issue.

Posted 1 year ago

Helo, David

The same thing happened to me; your blunt description was clear and accurate. I have retreated to v12.0.0

It shouldn't be too hard to put the build number or the date in the user portal. As it is now, we have to download something to see the build number.

WRI should use semantic versioning for Mathematica.

Thanks for this description. I am downloading the version after 6/23. It makes one wonder, however, the depth of testing before a release. This is such an "obvious" and easy to replicate bug in normal use that I wonder how it escaped through regression testing. Can the Wolfram folks assure us that testing is thorough before release, and that standards are not slipping? Mathematica is such an amazing tool, I would hate to see things rushed out.

Wolfram typical does not use beta testers for 'minor' updates. Perhaps they should.

I have been participating in beta tests since version 3. It is hard to catch everything, especially since the list of things to test for primarily consists of new functionality. However, this incident gives me a clue for the next beta test (12.2?, 13? or if they have Triskaidekaphobia, 14?)

I would imagine that many of the changes will be to ensure compatibility with macOS Big Sur and the release next year of Apple Silicon Macs, plus the usual bells and whistles. I would settle for fewer bugs and better integration with my hardware (I use macOS).

Posted 1 year ago

I must admit to being a bit surprised that my initial question still garners attention. As a veteran of first silicon validation for 386, 486 and Pentium plus three IPO's I have made my share of mistakes. I recall being embarrassed about delivering bugs not caught in Beta, also.

My major worry initially was that macOS was being depreciated. I have seen "mobile first" (e.g. Alpha?) strategies lead app suppliers to diverge from my preferred path several times and the architectural changes Apple announced recently must certainly cause suppliers to reconsider their priorities.

It looks like Wolfram continues to support the Mac and we should expect some bumps in the road as hardware and software environments change substantially... something I should remember.

I share your concern about macOS support. It's no secret that Theo Gray was largely responsible for continued Mac support during the 'dark days', and the macOS seems to be getting a bit less love since he has moved on.

I did find an issue with Big Sur that indicated that the modernization of the front end is incomplete. This problem, while not trivial, is easy to fix. However, I will look with great interest about the transition to Apple Silicon. As you may recall, Wolfram (and Theo Gray) were featured in the transition to Intel, so I had no doubts that Mathematica would be ready for the transition.

My main concern is not with basic functionality, which, I am sure, a recompile with a magic switch will take care of. Mathematica currently does not make full use of Apple GPU support, which makes it harder to accelerate Neural nets and other technologies. With the switch to Apple Silicon, we will undoubtedly get custom GPUs. These will be (probably) much better, but will require the use of Apple's Frameworks to work properly. Wolfram lagged behind support of Metal for graphics, and I fear that they will also all behind in support of Apple's other frameworks. Open source is all well and good, but even the best supported open source projects can't compete with the resources Apple has.

It will be interesting. I have a newish MacBook Pro that will get me through the transition, but I do hope to get a new MacBook sometime in the next year or so.

My feeling is that with more user feedback, Wolfram may be persuaded to adjust their priorities.

Interesting times....

Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
or Discard

Group Abstract Group Abstract