Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread Ramana Kumar
Hi David, I just wanted to add that the latest git master (26c678b6) still fails to build HOL4 with the following: Fail "Exception- InternalError: codeToPRegRev raised while compiling" similar to as we have discussed on other threads. Cheers, Ramana On 14 November 2017 at 09:32, Rob Arthan wro

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread Rob Arthan
David, > On 13 Nov 2017, at 20:47, David Matthews > wrote: > > Rob, > Thank you for putting up with this. No problem. Thank you for all your hard work with Poly/ML. > We're gradually making progress. I've pushed a further change and this > example now seems to work. I didn't actually get t

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread David Matthews
Actually, no. Each cell has a length word and pointers always point after this so the assertion check is correct. However the bug was related to there being a zero-sized cell, i.e. a cell that only has a length word, at the top of a segment. The cell address looks as though it is at the star

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread David Matthews
Rob, Thank you for putting up with this. We're gradually making progress. I've pushed a further change and this example now seems to work. I didn't actually get that assertion fault but something similar. Regards, David On 13/11/2017 16:32, Rob Arthan wrote: David, Thanks again, but I've

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread James Clarke
Without having looked at the code in question, that assertion seems somewhat unusual; a typical in-range check has >= and <. Perhaps it's a bit naive, but have you tried that? James > On 13 Nov 2017, at 16:32, Rob Arthan wrote: > > David, > > Thanks again, but I've got two ProofPower source

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread Rob Arthan
David, Thanks again, but I've got two ProofPower source files further on and then I get a different assertion failure: Assertion failed: (val.AsAddress() > descr->originalAddress && val.AsAddress() <= (char*)descr->originalAddress + descr->segmentSize), function RelocateAddressAt, file savesta

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread David Matthews
Thanks, both of you for your contributions. I've had another look at it and I've applied another fix. The problem was really that it was reading beyond the end of an array which meant that whether and how it failed depended on the values it found. Hopefully the latest fix (e968c38) will solv

Re: [polyml] Approaching release of 5.7.1

2017-11-13 Thread Phil Clayton
David, I also get a failure building ProofPower but not the same as Rob: pp-ml: savestate.cpp:881: void LoadRelocate::AddTreeRange(SpaceBTree**, unsigned int, uintptr_t, uintptr_t): Assertion `s >= r && s <= 256' failed. This is on a Linux x86_64 machine and occurs with commit 524fe72 (I hav