Sylvain BERTRAND wrote: > It's a bit on the side of the topic but GTK+ is now hard dependent through > pango on harfbuzz, a c++ component (object oriented cluster f....), which > pulls a dependency on c++ compiler/runtime.
WebKitGTK is also written in C++ so we need a C++ compiler and runtime anyway. > To safeguard netsurf from the toxic c++, a good thing would be to be able to > compile the entire user level stack of netsurf with a C compiler like tinycc. I am confused. Are you talking about NetSurf [0] or surf [1]? > Another way to see thing would be to go to EFL. After discussing that with the > EFL developer in charge, the EFL text layout engine can be set to use harfbuzz > or a internal basic one. But the developer strongly suggested that it may > become hard dependant on harfbuzz in the end. Then better stick with GTK+, > since at least it's protected with a GNU (L)GPL license, then GTK+3 support > would make sense. I don't see how EFL can be used as or with a website layout engine. Could you elaborate on that? > Since a unicode text layout engine is very important for a international > browser, and since the C toolkits are delegating that task to a c++ > component... what to do? Because, netsurf being the only C coded javascript > www browser out there, in my humble opinion, it's important to safeguard this > feature. The NetSurf I mentioned above actually only has experimental js support. > Personnally, I did a very partial C port of harfbuzz in C, > https://code.google.com/p/charfbuzz/ , which is a drop in remplacement of > harfbuzz (C api). I already mentioned it several times on this mailing list. > I'm not working on it right now (I'm on an alternative linux radeon driver), > but I will probably come back to unroll the original c++ code in C in the > futur. I don't think a port to C would work out well. If the code is as bad as you claim, a clean rewrite should take less time and lead to a better result. --Markus