In the release notes of this preview version, I saw quite some issues addressed that I am interested in.
Therefore, I tested a selection of points from the release notes. (This is an arbitrary selection, no structured test…)
Here are my findings:
Modules: see separate posts
Improved Code tab: works
Improved Info tab, LaTeX: the most useful application I had in mind is formulas; I tried $n$, but it did not show as a LaTeX formula; there is no other documentation yet
The automatic models update service: it works fine for models from NetLogo 5.0 and later, but not for earlier versions (I still have a bunch of 4.1 models around, and conversion then needs 5.x)
BehaviorSpace improvements, including error messages: it all works nicely (thanks!), except for one glitch when I allocated 8MB of memory (for 1MB it said it needed more)
static HTML files: yes! (needs some more updates to 7.1 though; e.g. no documentation on new BehaviorSpace options yet)
Copying text from Monitor widgets: nice!
Command Center improvements: mostly work, but recalling multi-line inputs should at least show a space for any newline
Improved command-line interface: no documentation found
Models Library dialog: nice!
about .nls files: when renaming to an invalid filename through the interface, I got an error message, but changing the name inside the __includes clause was accepted; I could even open the “invalid” file in the editor
reporting an error from the “internal error” dialog: it hangs when there is no Internet connection
check on names of globals: nice!
Auto-converter within NetLogo Desktop: it looks like this is not “within”; most points are pre-6.0 and then it refers to the external model-converter service; pre-5.0 it still breaks on an unsupported file-format (like 6.x did)
So, mostly good, with some small issues. If I find larger issues, I will report them separately.
Nice job!
Thank you for the extensive comments. The documentation for the next release is now updated with the missing items you mentioned. Regarding the other issues you raised:
Info tab LaTeX: $n$ works for me. Could you send an example model where it doesn’t work for you?
Auto-converter service: Unfortunately, there are some technical issues preventing us from supporting older versions of NetLogo in the automated environment, so in some cases you’ll need to do the first step manually. But there are quite a few bugs in older versions of NetLogo, so there’s a chance that your conversion just failed due to one of these bugs. If you send a more specific bug report elsewhere, we can take a look and see if it’s a solvable problem.
BehaviorSpace memory: You most definitely need more than 8MB. At least for me, just opening the app without even opening a model uses around 500MB. To prevent issues with platform differences we haven’t set a hard minimum, but something above 500MB is most likely required in all cases. I’ll make sure this is made clear in the BehaviorSpace documentation, and maybe in the dialog itself as well.
Command Center multi-line: I don’t have any issues with this on my end, pressing up arrow or selecting an item from the history dropdown does preserve the newlines. Could you send a screenshot or a more specific example of where it didn’t work for you?
.nls file names: What exactly was the “invalid” file name you used? If I do something like __includes [ “Untitled.nlogox” ], it throws an error, even if that model file exists in the same directory.
Error dialog: Good catch, it no longer hangs if there is no internet connection.
Desktop auto-converter: The reason for creating the separate auto-converter service was exactly this; it’s not feasible to support the loading of really old models in the latest version.
Well, I didn’t expect 8MB to be enough, either. I just wanted to see what happens. On 1MB you get an informative message, on 8MB it crashes with no clue (no OutOfMemoryException either). On 40MB it did work in my case (small model).
I did not even think of anything except .nls, but I did use non-identifier characters before .nls. Renaming by the rename functionality was not accepted, but changing the name inside __includes does work. Another example is __includes ["1E10.nls"] which is accepted if that file exists.
Ok, I see what you mean now. Although 1E10 would be an invalid module name, there are no identifier restrictions for .nls files. So I think the bug here is actually in the rename functionality; if you rename an .nlm file it should require a valid identifier as the filename, but not if you rename an .nls file.
My best guess as to what’s happening here is that NetLogo failed to fetch the script that renders the final LaTeX output, which is currently downloaded from a remote location when you launch the app. When you took the screenshot that you sent here, were you disconnected from the internet?
Regarding the CC issue: Thank you for the screenshot, I see what you mean now. We’ll fix that for the next release.
Thank you for the explanation.
Indeed, for testing software I use a stand-alone computer. I did not expect this functionality would need an Internet connection.
Now, I used my main machine for testing some issues that needed Internet connectivity.
The LaTeX formula works as expected.
For reporting an error from the error dialog, it does not hang.
But it does not give any further information either. Did it send the error?
And then, the message in the first dialog indicates to include the model. The second dialog does not ask for the model. It should at least ask confirmation that you want to send the model, if it does so automatically.
Do you expect the user to fill in their name? I did, but without e-mail address.
For me, I would still prefer copying the message and sending an e-mail or posting the issue on this forum.
We could display a confirmation popup once the error report is sent successfully, but then it would have the same issue that you originally raised; it would block while waiting for a response from the server. I think how it is now is the expected behavior, because the purpose of the report dialog in the app is slightly different than reporting something via email or the Forum. There’s no expectation that you would receive a response, just that you’re alerting the developers to a potential problem in case they are able to fix it. If you care about receiving a response or if you have a more abstract issue, you are still welcome to communicate on the Forum as you have been doing; the report dialog was just added to help users who want to quickly report an error as it happens but aren’t familiar with GitHub or the Forum.
I don’t think any personal information is expected, and the model is not sent automatically in any situation, because both would be privacy concerns. We can definitely make that more clear in the dialog itself.
