Skip to main content

SciDraw error loading package


I am running Ma 10.3. When I try to Get SciDraw package I get the following error:


AppendTo[$Path, "C:\\Users\\username\\mathematica"];
Get["SciDraw`"]


SetDelayed::write: "Tag MissingQ in MissingQ[expr_] is Protected".

Can someone confirm the problem?


Update


The Mathematicas v10-v10.3 contain bugs regardinng using a custom function for ticks and maybe this problem is inter-related. In v10.4, one can use a custom function again but for log scales, wolfram changed syntax so that it sends min and max values scaled rather than as in v9, however, in some cases, mathematica still sends unscaled values (FrameTicks or a composed function). See this post.



Answer



Unfortunately, SciDraw does indeed cause error messages because SciDraw tries to define functions which have been introduced into the core Wolfram language since SciDraw was introduced. One of these probably started in v10.2, over MissingQ, and the other probably dates to v10.4, over BoundingRegion. Users with earlier versions of Mathematica are unlikely to encounter these issues.


MissingQ, v10.2 and later


This is the error you found. Some digging around locates the following definition in FigData.m,



MissingQ[expr_] := ! FreeQ[expr, Missing];

with some text notes on the corresponding section of FigData.nb that indicate that this section is still relatively rough around the edges. This defines a pretty reasonable function, to check whether expr is 'missing', in the formal sense that it includes objects of the form Missing["reason"].


Unfortunately, a very similar function of the same name, MissingQ, was introduced in v10.2, with a slightly narrower meaning:



MissingQ[expr] gives True if expr has head Missing.



That is, MissingQ[2 Missing[]] will return True in the SciDraw implementation and False in the core implementation.


However, as far as I can tell, SciDraw's MissingQ is not used anywhere else in the code - indeed, it is not assigned a usage message, which is a good flag that it's not meant to be used yet - so if the message bothers you, it's probably safe to simply comment that line out. You're unlikely to break anything that isn't already broken (the error message indicates that the intended (re)definition of the function failed anyway.


BoundingRegion, v10.4 and later



As pointed out in the comments, there's a second error message - actually a pair of them - in later versions of Mathematica:


SetDelayed::write: Tag BoundingRegion in BoundingRegion[ObjectList:{(_Object?(And[
  <<2>>]&)|_?(And[<<2>>]&)|_Object?(And[<<2>>]&)|_?(And[<<2>>]&))..}] is Protected.

SetDelayed::write: Tag BoundingRegion in BoundingRegion[obj:_Object?(ObjectExistsQ[
  Slot[<<1>>]]&&MemberQ[ClassAncestry[<<1>>],FigAnchor]&)|_?(ObjectExistsQ[Object[
  <<1>>]]&&MemberQ[ClassAncestry[<<1>>],FigAnchor]&)|_Object?(ObjectExistsQ[Slot[
  <<1>>]]&&MemberQ[ClassAncestry[<<1>>],FigObject]&)|_?(ObjectExistsQ[Object[
  <<1>>]]&&MemberQ[ClassAncestry[<<1>>],FigObject]&)] is Protected.


The problem is exactly the same one, caused by the following definition in SciDraw's FigGeometry.m,


BoundingRegion[
  ObjectList:{(ObjectPattern[FigAnchor] | ObjectNamePattern[FigAnchor] | 
               ObjectPattern[FigObject] | ObjectNamePattern[FigObject] )..}
]:=Module[
    {},
    FigCheckInFigure[BoundingRegion];
    Canvas[ObjectCanvasBox[ObjectList]]
];


BoundingRegion[
  obj:((ObjectPattern[FigAnchor] | ObjectNamePattern[FigAnchor] | 
        ObjectPattern[FigObject] | ObjectNamePattern[FigObject] ))
]:=BoundingRegion[{obj}];

DeclareFigFallThroughError[BoundingRegion];

with the usage message



BoundingRegion[{p1,p2,…,obj1,obj2,…}] returns a region specification Canvas[{{xmin,xmax},{ymin,ymax}}] for the bounding box surrounding the given objects or points/anchors.




It is also documented in §7.4 of the SciDraw user guide, as



BoundingRegion[{p1,p2,…,obj1, obj2,…}]Returns a region specification which circumscribes the given points (or anchors) p1, p2, …, and objects obj1, obj2, …



The corresponding function, BoundingRegion, was introduced into the core language in v10.4, and it does do roughly the same sort of thing, but unfortunately the details are rather different:



BoundingRegion[{pt1, pt2,…}] gives the minimal axis-aligned bounding box for the points pt1, pt2, ….


BoundingRegion[{pt1, pt2,…},form] gives a bounding region of type form.


BoundingRegion[mesh,…] gives a bounding region for a MeshRegion or BoundaryMeshRegion.




This means that the two functions cannot really be substituted for each other. Fortunately, the situation is only slightly worse than with MissingQ:



  • BoundingRegion is not used elsewhere in the package as far as I can tell, so the conflict doesn't break anything else (and neither would removing the definition). It is defined as a user-accessible function, which is now lost. This will need to be addressed in future versions of the package.

  • Similarly to MissingQ, the error message indicates that the proposed (re)definition did not succeed, so if the error message bothers you, you can comment out those lines and the package's function will be identical.



This answer is current as of v11.0.0. Because of the nature of the functions defined in SciDraw, it's possible that further clashes will develop in the future - but hopefully not!


Comments

Post a Comment