Thank you Johathan; Didn't think of that. A novel and productive idea. There is only one 'rub', I had a linker (alas, not a loader error as originally stated) error which I reported and was told that it was difficult to change the linker and would probably not be done. This was, to me, the somewhat obscure "vtable undefined" message with an erroneous line reported where the error was to have occurred. Reporting the error to bugzilla would not be effective, in this case, because the error was well known and would not be fixed. An external explanation seems to be the only recourse. On the other hand, one error alone does not destroy a good idea. art ----- Original Message ---- From: Jonathan Wakely <jwakely.gcc@xxxxxxxxx> To: Arthur Schwarz <aschwarz1309@xxxxxxx> Cc: gcc-help@xxxxxxxxxxx Sent: Thu, July 12, 2012 2:54:42 AM Subject: Re: Help Message Manual On 11 July 2012 16:24, Arthur Schwarz wrote: > > Going forward, perhaps when a developer outputs a diagnostic message then > the developer provides a description of the message meaning to gcc(?). > Collecting these explanations becomes the manual? Or they just make the diagnostic self-explanatory. If the diagnostic isn't clear (to someone with a working knowledge of the language and compilation process) then it should probably be reported to bugzilla and improved.
