case_limit

[joerg-d]

For some of my projects I need a higher case_limit (e.g. 256). Is there any reason to set this to 64?

[reltham]

It is limited to 64, because that is what the original Spin compiler is limited to.

[konimaru]

It may be useful in the long run to introduce an option which explicitly removes certain limits.

[konimaru]

So this means the case limit still applies (64)? Just seemed odd that the limit was increased and then removed again (according to the log messages).

[reltham]

No, I had submitted a partial set of files, reverted that, and then submitted the full set of files. The partial set of files didn't have a summary description while the full set did, this may have caused your confusion? Just click the ... buttons to see the full description of the submits to see that the 3rd one also has the case_limit change to 256 in it.

[konimaru]

OK. It's Monday morning here before my first coffee. I'll check later. Thanks.

[konimaru]

Can you do something about the build errors (simple make invocation)?

g++ -Wall -g -static -o UnusedMethodUtils.o -c UnusedMethodUtils.cpp
UnusedMethodUtils.cpp: In function ‘int ScanLowerOpcode(unsigned char*, MethodUsage*, ObjectEntry*, unsigned char*)’:
UnusedMethodUtils.cpp:554:139: error: cast from ‘unsigned char*’ to ‘unsigned int’ loses precision [-fpermissive]
                 pUsage->pCalls[pUsage->nCurrCall].objaddress = (unsigned int)&pObject->pObject[pUsage->pCalls[pUsage->nCurrCall].objoffset];
                                                                                                                                           ^
UnusedMethodUtils.cpp:563:95: error: cast from ‘unsigned char*’ to ‘unsigned int’ loses precision [-fpermissive]
                 pUsage->pCalls[pUsage->nCurrCall].objaddress = (unsigned int)(pObject->pObject);
                                                                                               ^
UnusedMethodUtils.cpp: In function ‘void MarkCalls(MethodUsage*, ObjectEntry*)’:
UnusedMethodUtils.cpp:822:87: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
                 ObjectEntry* pSubObject = GetObject((unsigned char*)(pCall->objaddress));
                                                                                       ^
make[1]: *** [UnusedMethodUtils.o] Error 1

g++ (GCC) 4.8.3 20140911 (Red Hat 4.8.3-7)
Copyright (C) 2013 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

[reltham]

It compiles fine under gcc 4.6.2. Since when did they change it so that explicitly casting doesn't get rid of the "loses precision" thing? And when did that become an error instead of a warning?

[reltham]

I submitted a fixup for you that should work. I don't have gcc 4.8.3 available right now to test on, so if there were other problems besides what you listed then it'll have to wait. It's annoying how they keep changing what is a warning or error and that even with an explicit cast they now don't shut up about lost precision and call it an error.

[drawkula]

Compiles on Debian8 (gcc version 4.9.2).