www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - unfixed ICE issues

reply Benjamin Shropshire <ao pathlink.com> writes:
To help avoid duplicate issues, could the assert error messages in DMD be 
edited to include the issue number of bugs known to trigger them? That would 
be a little more robust than just line numbers and would also help people 
find work arounds.
Aug 12 2009
next sibling parent Don <nospam nospam.com> writes:
Benjamin Shropshire wrote:
 To help avoid duplicate issues, could the assert error messages in DMD 
 be edited to include the issue number of bugs known to trigger them? 
 That would be a little more robust than just line numbers and would also 
 help people find work arounds.
 

As of a few minutes ago, the number of ICE bugs without patches, which are common to D1 & D2 is down to just 3. with patches --- 15 unpatched D1&D --- 3 unpatched D1 only --- 10, + 2 which are Linux only unpatched D2 only --- 11, + 1 which is Linux only I suspect that 5 of the D2 bugs are duplicates. Many of them are newly discovered. So it should be a much less severe problem after the next release.
Aug 12 2009
prev sibling next sibling parent reply Stewart Gordon <smjg_1998 yahoo.com> writes:
Benjamin Shropshire wrote:
 To help avoid duplicate issues, could the assert error messages in DMD 
 be edited to include the issue number of bugs known to trigger them? 
 That would be a little more robust than just line numbers and would also 
 help people find work arounds.

I guess it would work in theory. How should we word it so that people don't immediately think "it's a known bug, so I don't need to report it" when it may actually be a very different cause of the same assert error? Moreover, would it report just the issue number, or the full URL of the bug report? Stewart.
Aug 12 2009
parent BCS <ao pathlink.com> writes:
Reply to Stewart,

 Benjamin Shropshire wrote:
 
 To help avoid duplicate issues, could the assert error messages in
 DMD be edited to include the issue number of bugs known to trigger
 them? That would be a little more robust than just line numbers and
 would also help people find work arounds.
 

don't immediately think "it's a known bug, so I don't need to report it" when it may actually be a very different cause of the same assert error? Moreover, would it report just the issue number, or the full URL of the bug report?

the short option would be to tack "(#3724)" or whatnot on the end. Alternately, it could be something like "(might be issue #3724)"
Aug 12 2009
prev sibling parent Stewart Gordon <smjg_1998 yahoo.com> writes:
Benjamin Shropshire wrote:
 To help avoid duplicate issues, could the assert error messages in DMD 
 be edited to include the issue number of bugs known to trigger them? 
 That would be a little more robust than just line numbers and would also 
 help people find work arounds.

We could use the same thing for forward reference bugs, especially those that have "forward reference" in the error message. Given that the presence of such an error message in the compiler code is nearly always an indication of a bug. Stewart.
Aug 17 2009