Why didn't Scinterm become one of the official Scintilla ports?

[rhaberkorn]

Hello Mitchell! I vaguely remember that there was some process going on to get Scinterm into the Scintilla repository itself and therefore to become one of the official ports. This would have the advantage of course that maintaining Scintilla wouldn't be your burden alone.

Why didn't this work out in the end?

[orbitalquark]

Hi Robin! I think it was because Neil didn't want the maintenance burden :)

A long time ago, I did maintain a LongTerm3 branch in the official repository. In exchange for being the maintainer, I was allowed to make Scinterm an official platform, but just for that branch. However, when my application transitioned to the 4.x and 5.x branches, Scinterm was separated into its own repository again.

Scinterm uses a few hacks to try and interpret what is being drawn. For example:

• scinterm/PlatCurses.cxx

  Lines 202 to 203 in fcc2c20

  	// Only called for CallTip arrows and INDIC_POINT[CHARACTER]. Assume the former. Line markers
  	// normally drawn as polygons are handled in `DrawLineMarker()`.

• scinterm/PlatCurses.cxx

  Lines 259 to 261 in fcc2c20

  	// Called to draw INDIC_ROUNDBOX and INDIC_STRAIGHTBOX indicators, text blobs, and translucent
  	// line states and selections.
  	// Does not draw INDIC_FULLBOX correctly.

• scinterm/PlatCurses.cxx

  Lines 287 to 288 in fcc2c20

  	// Only called when drawing indentation guides or during certain drawing operations when double
  	// buffering is enabled. Since the latter is not supported, assume the former.

• scinterm/PlatCurses.cxx

  Lines 344 to 346 in fcc2c20

  	// Called for drawing the caret, text blobs, and `MarkerSymbol::Character` line markers.
  	// When drawing control characters, *rc* needs to have its pixel padding removed since curses
  	// has smaller resolution. Similarly when drawing line markers, *rc* needs to be reshaped.

  
  This implicit behavior is brittle. More work needs to be done in Scintilla to make this kind of stuff explicit, and I haven't really bothered to try to develop and submit patches. It was difficult to get the necessary SurfaceImpl::DrawLineMarker(), SurfaceImpl::DrawWrapMarker() and SurfaceImpl::DrawTabArrow() methods approved.

[rhaberkorn]

Thanks for the clarification. I was just thinking today that a generic Cairo-based Scintilla backend could potentially be the base of almost all other backends (except Scinterm). It would also make it a lot easier to integrate Scintilla into other widget toolkits without an official Scintilla port. Especially considering that the GTK port is already Cairo-based. Perhaps I will pitch this idea to Neill.

Happy Easter holidays to you!