Resolved in 5f88aca3051928202c3882eff6f7b47892969d45

acceptance from @filipesaraiva is more important that mine, I think, because he have worked with the site. But, in my opinion, all looks right

Basic latex package on most of system haven't preview.sty, if I am not mistaken. For example, in Ubuntu, this is package preview-latex-style.
So, @asemke, you should add the embedded sty file and restore code for using embeded version.

Well, I hope, that this is change won't bring us problems

But what about older versions? For example, user can use Cantor 19.12 with sage 8.1 (yes, retrograde, but this is possible). I think, we should add some checking and use something like

...
else version < 8.9
    exit 1

so, we won't miss situations, when we don't found sage-env for versions, where this is needed.

In D24006#546359, @asemke wrote:

In D24006#533156, @sirgienko wrote:

Even though we have only one single entry to render in testMathRender(), there are multiple render tasks triggered here. We need to understand this. But again, this all seems to be an issue somewhere in MathRender, MathRenderTask and/or WorksheetTest.

I think, there isn't need to edit *.ipynb, because they are just an example input data, and also used in tests. So, I think, better revert the .ipynb changes, or update tests content testworksheet.cpp too.

Add explanation, how setKeywordList can be used

Anything but single-click initial setup, reasonable download and initial build times, less than 5 second compilation and launch times, and instant access to source code and documentation of any class or function (even those in external libraries) should be considered unacceptable.

How you want to implement this? 5 second compilation looks like something *physically impossible*, because of different computers have significantly different compilation possibility and different programs have a significantly different code amount.

Mm, any changes?

Add test and change the function description

Bugfix

In D23061#509851, @cullmann wrote:

Hmm, I think we want an unit test for this, too.
And I think it won't work like it is, before the call to initLookupForCaseSensitivity one needs to clear both m_keywordsSortedCaseSensitive and m_keywordsSortedCaseInsensitive vectors.

Use setKeywordList instead previous API

In D23061#509540, @cullmann wrote:

Actually, wouldn't it make more sense to just provide a way to set the full keyword list?
You can get the QStringlist will all keywords already with keywordList, a setKeywordList(...) would make more sense to me than this modifiers.
That would allow to alter it like you want and then write it back in one go with just one re-init of the mapping.