This is a suggestion by @paskevyc to use the same process as why3doc for detecting an ident.
why3doc uses a table from location to ident to know at which location each ident is defined: this is saved during parsing/typing using the flag
This MR generalizes the use of
Glob so that it is used with why3ide too: the flag is now set in
A notable change in Glob is the use of nested hashtbls: one hashtbl per file. This allows us to erase an hashtbl contents when
reading a file again (it avoids inconsistent locations see first commit).