muparser: avoid uncaught exception on invalid variable name #11793
Conversation
| } | ||
| catch (const mu::Parser::exception_type &) | ||
| { | ||
| m_bCompileFailed = true; |
There was a problem hiding this comment.
no interesting error message coming from mu::Parser we could CPLError() ?
There was a problem hiding this comment.
The message in the exception says that the variable name is invalid but doesn't say what the name was. Still, I've moved the site of the CPLError call to print an improved message.
There was a problem hiding this comment.
Not to be addressed by this PR necessarily, but I'm wondering about the actual benefits of passing std::string_view as argument if we end up creating a std::string. This is slightly more verbose in term of source code, an probably also a bit in term of binary size ?
There was a problem hiding this comment.
Actual benefit of std::string_view is negative, as you suggest. I think the longer-term improvement would be for CPLDebug and friends to accept a std::string_view argument instead of requiring everything be a C string. I haven't looked into it at all though, maybe there's some reason that won't work. Or maybe we should hold out for a couple of years and switch to std::format.
dbaston
force-pushed
the
vrt-muparser-fix
branch
from
0656f5c to
a8b75d0
Compare
No description provided.