[Wireshark-dev] invalid output in pidl generated dcercp-mapi & dercp-nspi dissectors

2016-02-11 Thread Aurélien Aptel
Hi, I'm in the process of syncing tools/pidl with samba (almost done, more on this later). Using master (0cd0844) bundled pidl, I wanted to make sure none of the generated dissectors were edited by hand so I've run the following (from epan/dissectors/pidl/README) to regenerate them: cd epan/

Re: [Wireshark-dev] Fix signed overflow

2016-02-11 Thread Guy Harris
Not that the variables should be signed in any case: https://code.wireshark.org/review/13897 (and not that we're likely to have 2^31-1 or 2^32-1 files to merge, so this is unlikely to be an issue in practice). ___ Se

Re: [Wireshark-dev] Fix signed overflow

2016-02-11 Thread Kevin Cox
On Wed, 2016-02-10 at 10:46 -0500, Michael McConville wrote: > Signed overflow is undefined, so the overflow check below is > technically > meaningless. Because we're only checking whether an increment will > overflow, we can compare j to G_MAXINT instead. > This is an important fix. The compiler

Re: [Wireshark-dev] Fix signed overflow

2016-02-11 Thread Michael Mann
Your patch has already been included: https://code.wireshark.org/review/13896 -Original Message- From: Michael McConville To: wireshark-dev Sent: Thu, Feb 11, 2016 10:24 am Subject: [Wireshark-dev] Fix signed overflow Signed overflow is undefined, so the overflow check below is te

[Wireshark-dev] Fix signed overflow

2016-02-11 Thread Michael McConville
Signed overflow is undefined, so the overflow check below is technically meaningless. Because we're only checking whether an increment will overflow, we can compare j to G_MAXINT instead. Thanks for your time, Michael diff --git a/wiretap/merge.c b/wiretap/merge.c index 19c11cd..06dc5db 100644 -