Hello all, My partner was just debugging a project that had tried to call a function without arguments in the code but the declaration wasn't declared with a void parameter list. It failed with an assertion that something was trying to ++ past the end of an ilist. I seem to remember Chris Lattner saying when he made the hand written AsmParser that it wasn't intended to be very robust but, searching the list, I can't find the post where he said it. That being said, is this worthy of a bug report? An error message from the AsmParser would be preferred over a seemingly unrelated assert. Cheers, --Sam Crow
On Nov 13, 2009, at 10:16 AM, Samuel Crow wrote:> Hello all, > > My partner was just debugging a project that had tried to call a > function without arguments in the code but the declaration wasn't > declared with a void parameter list. It failed with an assertion > that something was trying to ++ past the end of an ilist. > > I seem to remember Chris Lattner saying when he made the hand > written AsmParser that it wasn't intended to be very robust but, > searching the list, I can't find the post where he said it. That > being said, is this worthy of a bug report? An error message from > the AsmParser would be preferred over a seemingly unrelated assert.Please file a bug, it certainly should be "robust". Switching it to a hand written parser is intended to improve the robustness, not reduce it. In any case, this doesn't have anything to do with the parser, it sounds like the code is grammatically correct but semantically invalid. -Chris
...and forwarded to list. :-/ ----- Forwarded Message ----> From: Samuel Crow <samuraileumas at yahoo.com> > To: Chris Lattner <clattner at apple.com> > Sent: Fri, November 13, 2009 4:28:59 PM > Subject: Re: [LLVMdev] AsmParser is not robust > > Bug report #5486 filed. > > > > ----- Original Message ---- > > From: Chris Lattner > > To: Samuel Crow > > Cc: LLVM Developers Mailing List > > Sent: Fri, November 13, 2009 12:40:35 PM > > Subject: Re: [LLVMdev] AsmParser is not robust > > > > On Nov 13, 2009, at 10:16 AM, Samuel Crow wrote: > > > Hello all, > > > > > > My partner was just debugging a project that had tried to call a function > > without arguments in the code but the declaration wasn't declared with a void > > parameter list. It failed with an assertion that something was trying to ++ > > past the end of an ilist. > > > > > > I seem to remember Chris Lattner saying when he made the hand written > > AsmParser that it wasn't intended to be very robust but, searching the list, I > > > can't find the post where he said it. That being said, is this worthy of a > bug > > report? An error message from the AsmParser would be preferred over a > seemingly > > unrelated assert. > > > > Please file a bug, it certainly should be "robust". Switching it to a hand > > written parser is intended to improve the robustness, not reduce it. In any > > case, this doesn't have anything to do with the parser, it sounds like the > code > > is grammatically correct but semantically invalid. > > > > -Chris