<a href="/accounts/poupou">poupou</a> says: The <a href='http://www.ohloh.net/projects/mono'>mono</a> wiki has been updated with <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> 2.6 documentation.tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: When you look at log files for too long you always get headaches and, sometime, surprises. I was surprised by the memory requirements of the simple LINQ query that <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> does on defects. It turns out that ordering on enums is really bad (not tested if this only affects <a href='http://www.ohloh.net/projects/Mono'>Mono</a>) but casting the value to an int removes the allocations (good hint ;-). Previous baseline falls to 77291KB (-685KB)tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: <a href='http://www.ohloh.net/accounts/jbevain'>jbevain</a> approved my Has{x} patch to <a href='http://www.ohloh.net/projects/Cecil'>Cecil</a>. That will make it possible, once <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> rules are "fixed", to avoid the creation of (a lot of) empty collections. E.g. Custom attributes on fields are rare (about 2% on <a href='http://www.ohloh.net/projects/Mono'>Mono</a> System.Drawing.dll) but because Gendarme "touch" every CustomAttributes properties (not just on fields) all (100%) of them are created.tags: <a href="http://www.ohloh.net/accounts/jbevain">jbevain</a>,<a href="http://www.ohloh.net/p/cecil">Cecil</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: I started committing some <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> optimizations to HEAD tonight. They will soon be backported to the 2-2 branch. From last weekend profiling work I was able to reduce the memory required to analyze System.Drawing.dll (from <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 profile) from 154207KB (first run) to 127015KB (21st run), a reduction of 26MB (17.6%). Timings soon on my "official" blog.tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Late last night I uploaded the first preview (binaries and Win32 installer) for <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> 2.2. Documentation on <a href='http://www.ohloh.net/projects/Mono'>Mono</a> wiki has also been updated (and I need to look at automating this step). Benchmarking is next (while waiting for feedback) but I got quite a bit of patch to apply to HEAD first!tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2-2 branch was made earlier today. Setup is complete and <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> builds and test ok from the branch. I'll be moving my build-n-test activities to Windows and, once ready (the Makefile/MD/VS integration is still untested on Windows), build a preview of the 2.2 win32 installer.tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: The <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2-2 branch delay disrupted my plans a bit - but it did give me some free evenings to watch more PDC videos. I liked TL51 (Contracts and Pex) quite a bit since they are different yet quite near what tools, like <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>, and libraries, like <a href='http://www.ohloh.net/projects/Cecil'>Cecil</a> provides. I'm pretty sure you'll see those names mixed again :-)tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/cecil">Cecil</a> <a href="/accounts/poupou">poupou</a> says: I finally tested CheckParametersNullityInVisibleMethodsRule against <a href='http://www.ohloh.net/projects/Mono'>Mono</a> class libraries (mainly System.Security and System.Drawing) and felt confident enough (near zero false positives) to commit it in SVN in time for <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> 2.2. It's likely short-lived because Jesse is actively working on fixing NullDerefRule bugs and false positives - but will give users something useful *right now*.tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: While testing a (unrelated) patch I found a 64bits specific bug in <a href='http://www.ohloh.net/projects/Mono'>Mono</a> PrintingServicesUnix that <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> missed. So I extended <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> DoNotCastIntPtrToInt32Rule to cover the cases where Marshal.ReadInt32 (or ReadInt64) is used to read a pointer (i.e. casted back to [U]IntPtr) and renamed the rule to DoNotAssumeIntPtrSizeRule so it more accurately named with what it (now) does.tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: Time before the next <a href='http://www.ohloh.net/projects/Mono'>Mono</a> branch is getting short (November 3rd) so I started cleaning up the little bits I have lying around in <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> (and across 3 computers). That includes smaller rules like PreferEventsOverMethodsRule (committed tonight) and fixing defects found with rules still in development (and hopefully completed before the branch).tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: Finally committed the two rules I talked about two weeks ago: <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>.Rules.Correctness ReviewCastOnIntegerDivisionRule and ReviewCastOnIntegerMultiplicationRule. Both rules checks on cast around multiplication and division for common mistakes that returns unexpected values. They have been used to find a few bugs inside <a href='http://www.ohloh.net/projects/Mono'>Mono</a> class libraries.tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Finally completed the documentation for <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>.Rules.Design. Looks like it got a few typos to fix but the results on the new <a href='http://www.ohloh.net/projects/Mono'>Mono</a> web site just looks AWESOME. Next on the list are the performance rules... or maybe I should decide myself and commit my two rules, since they are already documented on the web site (but they are not in Design ;-)tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Rejoice! <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 was released today!!! and so was <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> 2.0 :-) Gendarme Binaries and Win32 packages are available on Ohloh @ <a href='https://www.ohloh.net/projects/gendarme/download'>https://www.ohloh.net/projects/gendarme/download</a> (note: identical to RC3 packages) while Linux packages (mono-tools) will soon be available from the Mono project (updated!) web site and/or from openSUSE build service.tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/jbevain">jbevain</a> says: <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 has been released!tags: <a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: I've been tuning two new <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> rules this weekend on <a href='http://www.ohloh.net/projects/Mono'>Mono</a> class library source code. A few more tweaks and it should land in SVN.tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Added new DoNotUseGetInterfaceToCheckAssignabilityRule to <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> BadPractice rules. The rule comes from a discussion between <a href='http://www.ohloh.net/accounts/jb'>jb</a> and <a href='http://www.ohloh.net/accounts/robertj'>robertj</a> on <a href='http://www.ohloh.net/projects/Mono'>Mono</a> leading to a patch to <a href='http://www.ohloh.net/projects/Moonlight'>Moonlight</a>. Then continued documenting the Correctness rules. Refactored FloatingComparisonRule (now renamed AvoidFloatingPointEqualityRule) and DoNotCompareWithNaNRule to share the same (engine-enabled ;-) base class.tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/moonlight">Moonlight</a> <a href="/accounts/poupou">poupou</a> says: While engine-ing some <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> rule I simply could not resist and took some time to remove allocations from AvoidUnnecessarySpecializationRule. That cut another 2.5MB of memory to analyze all <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 class libraries. Other smaller fixes were also backported to the 2.0 branch.tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/jason">jason</a> says: congrats to <a href='http://www.ohloh.net/projects/mono'>mono</a> on their second-life rollout: <a href='http://blog.secondlife.com/2008/08/20/mono-launch/'>http://blog.secondlife.com/2008/08/20/mono-launch/</a>tags: <a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: At this stage much of <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> action is reflected into it's <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 branch - but there are some differences (we're not HEAD-less ;-). The second part of <a href='http://www.ohloh.net/accounts/Nestor'>Nestor</a> GSoC work (the new rules) happens on HEAD (2.1 that will become Gendarme 2.2). This morning I also added a new rule AvoidExtensionMethodOnSystemObjectRule inside a new assembly, Gendarme.Rules.Design.Linq. More coming in Hack Week 3!tags: <a href="http://www.ohloh.net/accounts/nestor">nestor</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 preview 2 was tagged earlier this morning. I'll start benchmarking <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> later tonight, then update the win32 installer for it's release later this week. Meanwhile the feedback-based enhancements will still continue to happen on both the 2.0 branch and SVN HEAD - probably until Hack Week 3 :)tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: Fixed a bunch of small issues in the last days (well nights ;-). The most important one is that I finally found out why the <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> Wizard did not report as much defects than the console runner.... it simply did not default to the same visibility check* (visible only instead of all). * This is a new option in the console runner (by <a href='http://www.ohloh.net/accounts/Nestor'>Nestor</a>) which is very useful to apply the rules to <a href='http://www.ohloh.net/projects/Mono'>Mono</a>.tags: <a href="http://www.ohloh.net/accounts/nestor">nestor</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Fixed some false positives from AvoidRepetitiveCastsRule that could happen with the return value of method calls. Made a few other commits to <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>, mostly self-test, and to <a href='http://www.ohloh.net/projects/Mono'>Mono</a> (for validating the rule changes :-)tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: <a href='http://www.ohloh.net/accounts/latexer'>latexer</a> provided a patch to fix (i.e. ignore) all delegates in <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> AvoidUnusedParameterRule (bnc 414541). I also reduced the rule memory allocations by 86077KB (12%) over all <a href='http://www.ohloh.net/projects/Mono'>Mono</a> 2.0 assemblies, but I "limited" it to checking the first 64 parameters of each method (any method with more parameters has other issues IMHO ;-) tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Fixed two issues in <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> AvoidUncalledPrivateCodeRule. First one is about (unreported) unused properties (bnc #410000). The second is another case with generics. I also found a bug in TreeView.Sort under Vista (MS code, not <a href='http://www.ohloh.net/projects/Mono'>Mono</a>) that leads the wizard into infinite recursion (sorting rules). The fix was much easier than finding the problem (since I wrongly blamed my async code).tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/jbevain">jbevain</a> says: after 6 months of hacking <a href='http://www.ohloh.net/projects/Mono'>Mono</a>, back to hacking Moonlight, managed code needs some serious lovetags: <a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: I'm slowly completing/cleaning up my incomplete rule set... <a href='http://www.ohloh.net/projects/Mono'>Mono</a> branch will happen soon. This time I added AvoidLargeNumberOfLocalVariablesRule and AvoidRefAndOutParametersRule to <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>.tags: <a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: Earlier today Zoltan, aka <a href='http://www.ohloh.net/accounts/vargaz'>vargaz</a>, made a nice commit to change a few Type.GetType(string), which requires reflection, into faster typeof(x). How many more are there in <a href='http://www.ohloh.net/projects/mono'>mono</a> ? 84 tonight (hopefully less tomorrow) according to <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> new rule: AvoidTypeGetTypeForConstantStringsRule.tags: <a href="http://www.ohloh.net/accounts/vargaz">vargaz</a>,<a href="http://www.ohloh.net/p/mono">Mono</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a> <a href="/accounts/poupou">poupou</a> says: I just added DoNotRoundIntegersRule to <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> and a few instruction rocks to go along. Good news: no defect was found inside <a href='http://www.ohloh.net/projects/Mono'>Mono</a> class libraries. Bad news: it rains everydaytags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Still rainy :-( Anyway I committed ReviewUselessControlFlowRule for <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>. This is another rule to check for typos (like an extra ';' after a condition) and a great way to find empty code blocks (like forgotten TODO ;-). There are a bit over 100 of them inside <a href='http://www.ohloh.net/projects/Mono'>Mono</a> right now...tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: Just added ReviewDoubleAssignmentRule to <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a>. This check for multiple assignment to the same variable or field like: x = x = 42; I also added the first rocks (a la <a href='http://www.ohloh.net/projects/Mono'>Mono</a>.Rocks) for Instruction. This should simplify a few existing, and upcoming, rules :-)tags: <a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/jbevain">jbevain</a> says: Went from ~100 failures in the gigantic Expression Tree test to 3. Linq will sure look good in <a href='http://www.ohloh.net/projects/mono'>mono</a> 2.0.tags: <a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/poupou">poupou</a> says: GSoC 2008 has started this week. I'll suggest to <a href='http://www.ohloh.net/accounts/Nestor'>Nestor</a>, who is applying <a href='http://www.ohloh.net/projects/Gendarme'>Gendarme</a> to <a href='http://www.ohloh.net/projects/Mono'>Mono</a>, to try this journal for his "big" commits - unless he prefers blogging all of them.tags: <a href="http://www.ohloh.net/accounts/nestor">nestor</a>,<a href="http://www.ohloh.net/p/gendarme">Gendarme</a>,<a href="http://www.ohloh.net/p/mono">Mono</a> <a href="/accounts/jbevain">jbevain</a> says: Hacking on expression trees in <a href='http://www.ohloh.net/projects/mono'>mono</a>tags: <a href="http://www.ohloh.net/p/mono">Mono</a>