On 07/31/2014 06:29 AM, Joseph S. Myers wrote: > On Thu, 31 Jul 2014, Chen Gang wrote: > >> On 07/31/2014 06:12 AM, Joseph S. Myers wrote: >>> On Thu, 24 Jul 2014, Chen Gang wrote: >>> >>>> strlen() will get string length excluding '\0', but strcpy() will append >>>> '\0' in the end, so need XNEWVEC additional byte, or cause memory over >>>> flow. >>> >>> OK assuming it passed regression testing (with ChangeLog entry as usual, >>> and you need to say what platform the patch was bootstrapped / regression >>> tested on). >>> >> >> Excuse me, I only find it by reading source code, not give a test (for >> me, this kind of patch welcomes the related test, but not mandatory). > > I don't believe this particular patch needs a new regression test added to > the testsuite. > > But you still need to meet all the usual patch requirements - run the GCC > testsuite before and after the patch, and verify that it does not > introduce any new failures, and say what platform you did that testing on. > Even "obvious" patches can have non-obvious typos, hence the need to run > the testsuite as a sanity check. > > (For some sorts of patches it's different - e.g. for a patch to the > manual, "make info html pdf" is useful testing, but a testsuite run > isn't.) >
Thank you for your valuable information. And I shall try. For me, this kind of tests are necessary for all kinds of patches (include this kind of patch). But excuse me, I am a newbie, I am not quit familiar with testsuite, so maybe I can not finish soon. But I should finish the related test with testsuite within this week end (within 2014-08-03). And still welcome any related ideas, suggestions, and completions. Thanks. Thanks. -- Chen Gang Open, share, and attitude like air, water, and life which God blessed
