You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Mar 8, 2020. It is now read-only.
Repository navigation
This repository was archived by the owner on Mar 8, 2020. It is now read-only.
After some deep investigation, we can submit another issue to CDT, but I have a feeling that cdt-core is almost frozen and nothing is going on there. I submitted a previous issue almost month ago, and nobody took a look, so far.
Clang is an obvious choice, but migrating to a different C++ parser is likely to be a fairly involved project. Is the scope of the failure right now something that affects a lot of programs, or is it something we can afford to mitigate in a less-expensive way for now?
@creachadair It looks like it happens only on (real) C++ parser test files, not on files written by humans, so it may be safe to ignore. Also, according to the issue description, there is only two files that produce this error in the PGA (or other dataset used for the large parse by ML).
Activity
kuba-- commentedon Sep 20, 2019
NegativeArraySizeExceptionmay come from stack overflow trying to annotate an extremely deep AST.:Example
Details
dennwc commentedon Sep 20, 2019
For the clarification, @kuba--'s example is from the file mentioned by the log.
Given that it indeed looks like stack overflow (or similar kind of issue), we may want to report it to CDT.
kuba-- commentedon Sep 20, 2019
After some deep investigation, we can submit another issue to CDT, but I have a feeling that
cdt-coreis almost frozen and nothing is going on there. I submitted a previous issue almost month ago, and nobody took a look, so far.dennwc commentedon Sep 20, 2019
If there is no activity on PRs we should consider either forking it, or maybe use something else as the C/C++ parser.
kuba-- commentedon Sep 22, 2019
switching to something else will require much more work, but fork sounds reasonable to me
creachadair commentedon Sep 23, 2019
Clang is an obvious choice, but migrating to a different C++ parser is likely to be a fairly involved project. Is the scope of the failure right now something that affects a lot of programs, or is it something we can afford to mitigate in a less-expensive way for now?
dennwc commentedon Sep 24, 2019
@creachadair It looks like it happens only on (real) C++ parser test files, not on files written by humans, so it may be safe to ignore. Also, according to the issue description, there is only two files that produce this error in the PGA (or other dataset used for the large parse by ML).
kuba-- commentedon Oct 1, 2019
More details about
CDT'sexception:kuba-- commentedon Oct 1, 2019
Shall we submit yet another issue to CDT and keep it blocked or just close it?
I agree, the reproducible example is pretty artificial.
[-]Error serializing the AST to JSON[/-][+]NegativeArraySizeException (StackOverflowError)[/+]dennwc commentedon Oct 1, 2019
Let's keep it blocked.