Re: [Quantlib-dev] Interest in fixing MSVC Level 4 warnings

[Luigi B. <lui...@gm...>]

Mike,
    thanks a lot, there's a lot of useful information in there. I'll
try and get at least some of those changes into next release. I'll get
back to you if I have questions.

Later,
    Luigi


On Thu, Feb 28, 2013 at 5:46 AM, Mike Sharpe <ma...@gm...> wrote:
> I apologize for the late follow up to looking into getting these warnings
> fixed, but I have a VC++ 2012 warning free fork of the github mirror at
> https://github.com/masrtis/quantlib. I've created a branch,
> VC11Level4WarningFixes, that contains the changes I've made.
>
> The vast majority of warnings were C4512: cannot generate an assignment
> operator for 'class'. This is because of the use of const members and
> references in classes. To silence these warnings as I was going through, I
> used boost::noncopyable and that seems to be a bit overkill. Ideally it
> would be nice to fix these warnings by adding a similar class that just
> disables the assignment operator for classes with const members and maybe
> keep using boost::noncopyable if a class has reference members, but I think
> the amount of files affected for the amount of benefit gained is pretty
> small. Because of this, I recommend not merging down the entire branch.
>
> I did, however, find some bugs and do some minor refactoring for other
> warnings. I've created tags on the branch that mark changes that should be
> made and also changes that would be good to get reviewed. A summary of these
> is below.
>
> FIX1: Disables the conditional expression is constant warning for the
> QL_FAIL macro using Microsoft specific pragmas that work correctly with
> macro expansion. It's worth considering because this change fixes 1,276
> warnings on its own.
>
> FIX2: Removes unreachable code from CPICouponPricer::optionletPriceImp. I
> wasn't sure if the code was left in for documentation purposes, but if so it
> should be commented out.
>
> FIX3: Disables a bunch of warnings that leaked from uBLAS headers.
>
> FIX4: Replaced a runtime check of a numeric template parameter with
> BOOST_STATIC_ASSERT to catch the programming error at compile time instead
> of runtime.
>
> FIX5: Fixed uninitialized variable in CPICapFloorTermPriceSurface.
>
> FIX6: Fixed uninitialized variable in CPICapFloor.
>
> FIX7: Refactored code in Concentrating1dMesher to improve variable scope.
>
> REVIEW1: I wasn't sure why RendistatoCalculator was using private
> inheritance from LazyObject. I fixed the warning by using
> boost::noncopyable, but I suspect there's a missing public keyword here.
>
> FIX8: Disables some more warnings from uBLAS.
>
> REVIEW2: There might need to be some code added to the check a return value
> in Garch11::calibrate_r2().
>
> FIX9: Remove a duplicated return statement from
> OptionletStripper::optionletStrikes().
>
> FIX10: Added a static_cast around a toupper() call in
> PeriodParser::parseOnePeriod() to silence a size conversion warning.
>
> REVIEW3, REVIEW4: There is a lot of commented out code in function bodies in
> experimental/credit/distribution.cpp and
> experimental/credit/lossdistribution.cpp. I ended up commenting out what
> appeared to be similar code that was actually active and therefore
> generating unreachable code warnings. It's worth reviewing these files to
> see if the commented out code is still needed.
>
> FIX11: Made sure that a const array declaration in
> math/randomnumbers/primitivepolynomials.h was always considered extern, even
> when it was included in the companion .c file.
>
> FIX12: Disabled a conditional expression is constant warning in
> ZigguratRng::nextGaussian().
>
> FIX13: Added files to the VC11 projects that were checked in after I had
> created my branch.
>
> I hope this information is useful. Let me know if there's any questions or
> if I can do anything to help get these changes integrated into quantlib.
>
> Mike
>
>
> On Mon, Jan 14, 2013 at 9:20 PM, Mike Sharpe <ma...@gm...> wrote:
>>
>> Hi Luigi,
>>
>> Thanks for your reply.
>>
>> A quick upper bound on the amount of warnings that would need to be fixed
>> is 1800. This is high because VS reports the same warning multiple times if
>> the code in question is in headers, and also because boost has some level 4
>> warnings in uBLAS headers.
>>
>> I'm more than willing to use my spare time to generate patches that would
>> fix all the warnings, especially if they appear in headers so people who
>> include quantlib headers can do so without needing to resort to workarounds
>> if they want to bump up the warning level. When I submit the patches, I'll
>> describe which warning the patch fixes and we can go from there.
>>
>> Look for patches over the next week or two.
>>
>> Mike
>>
>>
>> On Thu, Jan 10, 2013 at 9:31 AM, Luigi Ballabio <lui...@gm...>
>> wrote:
>>>
>>> Hi Mike,
>>>     given that we won't be able to go entirely warning-free (due to
>>> QL_FAIL etc.) I wouldn't want you to invest too much time on this.
>>> But if you want to have a look at the warnings and check if there's
>>> anything that should be fixed, by all means go ahead.  A count might
>>> indeed be useful in order to assess if the thing is worth pursuing.
>>>
>>> Later,
>>>     Luigi
>>>
>>>
>>>
>>> On Sun, Jan 6, 2013 at 11:35 PM, Mike Sharpe <ma...@gm...> wrote:
>>> > Hi all,
>>> >
>>> > First, I'd like to start off by saying thanks for your hard work on
>>> > quantlib! This is my first post to a quantlib mailing list. I've
>>> > recently
>>> > developed an interest in mathematical finance and discovered quantlib
>>> > over
>>> > the holidays, which has only piqued my interest more.
>>> >
>>> > As I'm still new to the codebase and most of the techniques that are
>>> > used in
>>> > the code, I'd like to give back in some other way and one of the ways I
>>> > can
>>> > do that is by looking into fixing some of the level 4 warnings for the
>>> > Microsoft compilers. Because the warnings could appear anywhere and I
>>> > don't
>>> > want to cause any difficulties for other work, I wanted to check here
>>> > to see
>>> > if there was any interest in seeing these warnings fixed.
>>> >
>>> > I'm not necessarily proposing fixing all of the warnings that I've seen
>>> > in
>>> > the code (QL_FAIL triggers C4127 "conditional expression is constant"
>>> > because of the do { } while(false) macro guard, for example), but I do
>>> > think
>>> > it'd a good idea to look through the list and fix some of them. If a
>>> > count
>>> > of warnings would help, I can follow up with that soon.
>>> >
>>> > Mike
>>> >
>>> >
>>> >
>>> > ------------------------------------------------------------------------------
>>> > Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
>>> > MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
>>> > with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
>>> > MVPs and experts. ON SALE this month only -- learn more at:
>>> > http://p.sf.net/sfu/learnmore_123012
>>> > _______________________________________________
>>> > QuantLib-dev mailing list
>>> > Qua...@li...
>>> > https://lists.sourceforge.net/lists/listinfo/quantlib-dev
>>> >
>>
>>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb
> _______________________________________________
> QuantLib-dev mailing list
> Qua...@li...
> https://lists.sourceforge.net/lists/listinfo/quantlib-dev
>