4 ms·
Trying to Sell PVS-Studio to Google, or New Bugs in Chromium
- taspeotis 13y agoThe Chromium developers disregard Visual Studio and don't use Makefile but still manage somehow to write an incredibly high-quality code. How come? I think the answer is because high-quality code doesn't come from Visual Studio and/or makefiles.
- yeukhon 13y agoNow you have just made an enemy of roughly half of the C/C++ developers in this world lol
- taspeotis 13y agoWhy? It doesn't matter what IDE you use. It doesn't write the code for you.
- Danieru 13y agoExcept everyone everywhere hates their build system. No one is making enemies here. No project thinks their code quality is a result of choosing make, or automake, or cmake, or qmake, or scons, or ant, or premake, or now I hear someone went and wrote a new one: GYP. Because of course if there is one thing the world needs, it's Yet Another Build System. Which as it happens, is also a build system: http://www.op59.net/yabs/readme.html http://www.op59.net/yabs/readme.html
- raverbashing 13y agoI have a personal corollary that I am now extending. Two things always suck: bug tracking and build systems. What I think happens is that even if the foundations are different (and better), the ideas on how to do it are still the same as the first build systems. And that's the problem Overcomplicated, unhelpful, trying to fix language problems. Yes Makefiles are bad but a handcrafted one (or minimally automated) is usually much cleaner than most of the automated stuff.
- deathanatos 13y agoHow? Visual Studio isn't available for my OS, and Makefiles just don't cut it anymore. (I typically use CMake, but I'll use anything that gives me fast build times and gets out of the way. Make typically isn't that.)¹ 1. I can do a build in one command. 2. It doesn't take forever. 3. It's easy to set up & modify. ¹Although, CMake generates makefiles. Does that count?
- krakensden 13y agoThe number I'm most interested in: the false positive count.
- carterschonwald 13y agowell said! The false positive rate of most static analysis tools is the biggest barrier to wide spread use!
- nl 13y agoThe PVS-Studio analyzer has generated 3582 first-level warnings (the set of General Analysis rules) for this project. This number is very big. Most of them are false positives. .. [a] simple operation lets us subtract 2300 false positives from the total number of messages (3528). We have at one instant eliminated 65% of messages. Now we don't have to waste time examining them. [1] http://www.viva64.com/en/b/0205/ http://www.viva64.com/en/b/0205/
- kodablah 13y agoThis was discussed in the August review [1]. Basically, they found ~3500 problems, and whittled it down to ~1200 after removing just one warning about a strangely implemented macro. This is similar to many other static analysis tools (e.g. findbugs)...you have to work the ruleset for your app sometimes. 1 - http://www.viva64.com/en/b/0205/#ID0EV4AO http://www.viva64.com/en/b/0205/#ID0EV4AO
- acqq 13y agoThe article is not easy to read, a human editor would help. Still note that the first found bug looks serious, it seems to lower the security: it's in ssl3_SendEncryptedExtensions and compares wrong.
- habr 13y agoДАИТИ ЭНВАИТ НА ХАБР!