To check whether an ExternalText could be written with Epsilon's fonts,
UTF8Helper made a reference to Kandinsky, which is prohibited. This
check is now done in Escher, before dispatching the event.
Change-Id: I55e9db1ba43c3115775499db47b90a6bdd7cc7b3
Change the way the clipboard is fetched, so as to preserve the
Poincare/Python translation feature, and remove the cap on the length of
text being imported into the simulator.
Change-Id: I0f3e83c0c8aa4b64eb08d882aa6891b31f191e22
Add methods to interface Epsilon's clipboard with the system's when
running a simulator.
Note that the clipboard still uses a buffer located in the
Escher::Clipboard class, as some features require a buffer to execute
computations on the clipboard's content.
Change-Id: I14c19615805d38735e64d481c617863db22db9bc
Added an event to represent the typing of text that is not on the
device's keyboard. ExternalText event's text is read from a buffer, that
will be filled when the event is generated.
ExternalText only exists on the simulator.
Change-Id: Ie78d2c7c2de91da986a1ce2130a5ecd123db48ee
When a formula is pasted in a script or in the shell, some symbols are
replaced by their equivalent in python :
x turns into *
^ turns into **
√ turns into sqrt
etc
Change-Id: If6f2a22d4f3c148c2655e0892023b0e28058a9a6
For instance, if the cursor is: "he|llo" and the user adds the letter
'i', there is no autocompletion provided as the cursor is still in the
middle of a word.
setjmp is not guaranteed to collect all registers without modification
on all platforms.
This fixes the following bug: when the pointer of a newly allocated
object on the Python heap is stored in rpb registers on x86_64 arch, it
was not collected by the garbarge collector.
Qualification of an integer return type as volatile is ignored by the
compiler. To clarify, an integer value (not a variable), which is what
the return type is, can not be assigned to anyway, so qualification as
volatile does not make sense.
The buffer length did not take into account the \n removal. In fact, we
could just remove the \n while also removing the empty code points,
which is what we do now.
Scenario: Copy a text with \n (for instance from a script), then paste
it in a script name -> if \n are replaed with \0 instead of just being
removed, there are problems with the extension
We check the context: if an identifier is already a function, we
interpret it as such, otherwise it is a symbol.
This way, in the Equation app, x(x+1) is interpreted as x*(x+1)
