Double click on MacOSX doesn't work
For some reason, on Macs, having a single click handler interferes with double-click handling. So I ended up bodgying a double-click handler by adding a timeout to the single click handler, and performing the double click action if a second click is received in the timeout period. It works, but now single-click handling is slightly laggy, so I've left the original double-click binding in place for non-Mac OSes. I also fixed up the missing help and description entries on notes.
Double click on MacOSX doesn't work
Minsky on Mac does not remember location of last file opened or saved.
Save on initial open doesn't save; Save dialog defaults to /
physicalMem implementation for MacOSX doesn't compile
Movie
Add row buttons malfunctioning in Edit mode
Double click on MacOSX doesn't work
MKY file
Text scaling error
Context sensisitive help open editor forsome items
It doesn't affect Macs either. It's a weird one. I'll place this in backlog.
This problem doesn't affect Windows.
Show values not working properly
Got it. You need to open the model file from the menu, not command line. displayValue/displayStyle needs to be restored to preferences on loading the model.
Zero tolerance on Godley
I altered the code for computing the sum of the stock variables to round to 0 when the relative error is less than 10E-4. I don't have the exact example above to truly test, but seems to work on some other examples I have here.
Show values not working properly
Maybe a Windows bug. I can't reproduce this on Linux. 2.28.0 build exhibits this issue. Will need to try latest build on Windows.
Canvas redraw not exception safe
Actually, the trigger for this ticket is a regression in the code that converts schema1 files into schema2. I've just come across this during release testing for Graeber.
Can't grab Godley Table in Edit mode when Variables displayed
After quite a bit of rework, I think we're good now.
Show Values in Godley Table should shrink display to fit inside current bounding box
Research and replace linkchecker for checking broken links in the documentation
Can install the latest linkchecker using pip3. Notes added into doc/makeDoc.sh
Research and replace linkchecker for checking broken links in the documentation
Can install the latest linkchecker using pip3. Notes added into doc/makeDoc.sh
Undeletable Godley Rows
The problem is caused by the Central bank and Treasury Godley tables having conflicting definitions of Repay_{GD}, Minsky cannot determine which sign you want - either is mathematically correct. Either change the Central Bank's entry to -Repay_{GD} or change the Treasury's entry to Repay_{GD} and then the cycle stops. I don't think this is a bug, but rather an unexpected feature of double entry bookkeeping.
Restore object size
Ticket moved from /p/minsky/tickets/1330/ Can't be converted: _priority: 3urgent
Perhaps, although such a command will always be buried in the menus, so unlikely to be deployed. Better would be for the autolayout command to do a better job - I will be doing some work on that in the next iteration, though not for Graeber, In the meantime, this should really be classified as a feature request in the backlog.
Plotwidgets not responding to mouse events
Working now...
Switch not switching visually
This was a regression. Fixed now.
Context sensisitive help open editor forsome items
Unbalanced Godley Table
Integral and UserFunctions edit dialog have extraneous "relative" checkbox
OK by me. I'll keep an eye out for it in future model building.
That's a blow. I actually disagree - we should resolve this as CantReproduce, as we have nothing to go on, and no reasonable chance that there's anything to do. We can always reopen it if something further comes to light, although its more likely we'll open a new ticket, as this will be forgotten about.
Nup, no luck. Backburner, I think.
Actually, I might have a recording--I'll check.
Unfortunately no! It just happened while I was designing the table: by the time I'd seen the bug, it was too late to record the steps that led to it. It might be best to put this on the backlog and wait till I reproduce it some other time.
Unbalanced Godley Table
I can from the .mky file that the RBNZ has the Treasury pays interest line but the Banks table does not. When opened, Minsky corrects this by deleting the unbalanced entry from the RBNZ table. You can then add it back in, and Minsky autopopulates the column in the Banks table. What I don't know is what steps you did to get it into this state. Any hints?
MKY file
Switch not switching visually
A Notes Tab
Data not displaying
Rearrange palette tools
Axis bug for you
Data problem in formula
Sort by value not working
ArgumentsNotConformal
Different colours for Ravel Axis text labels
Axes wrong way round
Axis Not Flipping
Wires disappear
Screwy date formatting on horizontal data import
Disappearing Axis
Appears to have been fixed between 2.21.0 (version when this ticket was raised) and now (version 2.29.0-beta.6)
Very slow performance
Lock Axes needs to enable axes with different names but same data to be locked
I fixed up the outstanding issues here, and did step 3, adding persistence to .rvl files.
Yeah, here we are again! I think we need three distinct operations. We shouldn't be forcing users to choose between them, dropping one to do another. The standard example is (a) sort by value; (b) select top 3. Sometimes a user will want to track 3 countries over time, but normally they'll want to see the top 3 at each time slice. So we need to make that a possibility; if it isn't we frustrate users, the same way that Tableau does (so Martin North uses Excel for some analysis of his Tableau data...
Comment on video: You need to decouple sort order to be able to independently pick slices. Pick slices is a type of "sort order". Maybe we can come up with a different name for the link sort order option. It's kind of "sorting , permutations and filtering", but that's too long a label. The aggregation issue is solved in https://sourceforge.net/p/minsky/ravel/181/
Aggregation error
Ravel file
Quickly fixed. Changed the logic to apply any custom ordering and/or calipers prior to aggregation.
Told you so! ;). They need to be distinct operations.
Told ya! ;)
I didn't realise - it's the interaction between "pick slices" and aggregation that is the issue.
Ravel file
The attached more clearly illustrates the bug. Both plots should be the same.
Unbalanced Godley Table
Aggregation error
StrangeAggregationLockAxesEnhancedForm20210831.mp4 https://drive.google.com/file/d/1tqccFWbESYTDyMmCBWChNrJJ75XyTQZV/view?usp=drive_web
Some weird effects going on for aggregation with this Ravel. Movie file in next message.
MKY File
I spoke too soon. Just generated a fatal crash with this beta. See video. RVL file in the next message.
Lock Axes needs to enable axes with different names but same data to be locked
I'll take a look at the exact issue in the movie tomorrow. Re incomensurate dates, what we want is for the slicer to jump to the nearest slice - applicable for time and value axes, of course. This will require some work within the Ravel codebase itself. But it will also depend on the proposed refactor of Ravel to make use of the Civita library - currently Ravel doesn't understand time or value - which of course is scheduled for after the Graeber release is done. Other finessing: when only two ravels...
Link to Movie: https://drive.google.com/file/d/1OXxW-SGNVbjheH7v1EyxSFHVQ0o_LW4H/view?usp=drive_web
Pretty damn good for a first pass. One error: all the dimensions are allocated to the first Ravel, rather than just the dimensions native to that Ravel. I'll post the movie in a sec. It will probably be too large for SourceForge, so I'll send by email first to create a Google Docs link. One other issue turned up here: how to handle dates when one series is Quarterly and the other Monthly. But that's something more fundamental than just this issue of locking Ravels.
I'll get you a build after each step. No promises of step 2 being complete by tomorrow night, though...
I've broken this task down to 1. Creating the popup dialog supporting this table, including any C++ code required to support getting the information 2. Modifying broadcastToLockGroup with the necessary logic to implement this finer-grained control 3. Modifying Schema3 to persist the lock group settings. I believe I can do this in a backward compatible way, so as to not need bumping the schema.
Agreed. Fortunately you write CSVs with a metadata row, which is an improvement over the wild version. We do need to import the CSV without hassles--though equally importing XLS/XLSX formats would get around some of our problems, but with much more work--worth doing only after we have funding. For the meantime, handling what Excel does to one of our CSV exports is critical until such time as we have an in-Ravel editing capability.
An interesting article, pertinent to this ticket https://www.bitsondisk.com/writing/2021/retire-the-csv/. We've got to support CSV for the interim, and write the Excel dialect. I followed the RFC, but Excel does something a bit different to escape quotes.
PS The four input/outputs are just a thought about the orientation of the compact Ravel--horizontal or vertical? We also need to support flipping and rotating Ravels. And maybe two input ports could be used for merging purposes.
Add compact display mode for Ravels
Everything loads fine. This bug is definitely dead!
Fatal crash on file load