Interestingly, the intuitive calls convert(...,sin) and convert(...,cos) are not documented...E.g. see also here.

I can confirm that the same effect about stopping at SubexpressionMenu is observable by using mdcs. I do not believe that the dotm length of a call to Typesetting:-Typeset of the expression has relevance here. In the normal computation flow of 2-D input, Typesetting:-Parse receives from the Standard GUI a TypeMK version of the input in the form of Typesetting member calls. And it returns to the Java code a pair of (a possibly modified) TypeMK and Inert expressions. Nowhere Typesetting:-Parse and family handles dotm format, as far as I have seen. And Java code generating the input for Typesetting:-Parse would have no need of it either as the C kernel does not intervene (after observation of the localhost TCP stream) in the process generating the popup from selected output.

I can confirm that the same effect about stopping at SubexpressionMenu is observable by using mdcs. I do not believe that the dotm length of a call to Typesetting:-Typeset of the expression has relevance here. In the normal computation flow of 2-D input, Typesetting:-Parse receives from the Standard GUI a TypeMK version of the input in the form of Typesetting member calls. And it returns to the Java code a pair of (a possibly modified) TypeMK and Inert expressions. Nowhere Typesetting:-Parse and family handles dotm format, as far as I have seen. And Java code generating the input for Typesetting:-Parse would have no need of it either as the C kernel does not intervene (after observation of the localhost TCP stream) in the process generating the popup from selected output.

@spradlig 

About some reasons and other GUI issues you may like to read this thread and linked ones therein.

@spradlig 

About some reasons and other GUI issues you may like to read this thread and linked ones therein.

A problem with style=point 3D plots is rotation. In Standard GUI, with a grid of [70,70], as used by Christopher above, rotation still goes smoothly in my machine. But already with [100,100], to get a nicer looking surface, rotation becomes jumpy. I find it remarkable as it is just about a factor 2 more points. For the same expression to be ploted, I do not observe a similar problem in ordinary style=patch 3D plots, even at higher grid densities.

@Carl Love 

The usage of preformatted mode, which does not wrap. You can see it typeset in monospace font (which one, I think, depends on what is configured in your browser) . This is good for code and output (though if lines are too long CRs have to be introduced manually), but not for ordinary text.

@Carl Love 

The hack for reading long lines is saving the page as a text file.

It makes me remember a request of mine to support, October 10 1997:

I would appreciate very much an option discont for plot3d.

The answer, four days later:

I will forward your suggestion for a discont setting in plot3d on to our developers to consider.

Just over 15 years ago...

In the mean time, Mathematica got it by adaptive mesh and points its finger at Maple with this same branch cut plot.

What is that?

@acer 

Yes I know, but apparently, the font size modifier only works with the Typesetting export calls.

@Carl Love 

Actually the italic font is selected by the mi export, so that you could write simply:

Typesetting:-mi("x",fontweight = "bold",mathcolor = "#c800c8",
size="18");

and get the same output. Similarly, the mo export produces output in roman (upright) font, even with the "italic" modifier:

Typesetting:-mo("x",fontstyle = "italic",fontweight = "bold",mathcolor = "#c800c8",
size="18");

No, I think that there is no useful reference documentation on the Typesetting package and TypeMK. Paulina has written here several times in the past telling that this is "work in progress", i.e. subject to change, so that they it will not be documented. So, along the years I have collected a somewhat large amount of "experimental" material on these subjects.

There are two reasons why I will not write a post on this subject in the short term. 1. Scarcity of time to organize this material into a readable form. 2. I am in "protest mode" (using my protest user name) since the current version of this forum was introduced in early June 2010, as it is a big disaster and the administrators do not show any serious intention of change (I have explained in detail my reasons several times, mainly shortly after that date, and occasionally since then). So, I am writing here only on a few specific subjects and for a few specific purposes. In particular, I am not writing "posts".

@Carl Love 

I would say that it is trivial by using the inertization prefix % (see ?value ):

%log10(x);
value(%);
                           %log10(x)
                             ln(x) 
                             ------
                             ln(10)

I would rephrase acer's second approach by means of the procedure dividend:

T := -z/(x-z):
dividend(T,sign(denom(T),z),[u->u,u->sort(u,[z])]);
                                       z
                                     -----
                                     z - x

dividend(T,sign(denom(T),z),[u->u,u->sort(u,[z],':-descending')]);
                                       z
                                     -----
                                     z - x

I would rephrase acer's second approach by means of the procedure dividend:

T := -z/(x-z):
dividend(T,sign(denom(T),z),[u->u,u->sort(u,[z])]);
                                       z
                                     -----
                                     z - x

dividend(T,sign(denom(T),z),[u->u,u->sort(u,[z],':-descending')]);
                                       z
                                     -----
                                     z - x