Re: [sumo-devel] =?iso-8859-15?q?Sorry=2Cmistake_=28was=3A_Re=3A_Ques?= =?iso-8859-15?q?tions_about_a_migration_to_SVN_=26_Release0=2E9=2E5=29?=

[<d.k...@we...>]

Hello!

> -----Urspr=FCngliche Nachricht-----
> Von: <Dan...@dl...>
> Gesendet: 15.11.06 11:58:24
> An: <d.k...@we...>
> Betreff: WG: [sumo-devel] Sorry,mistake (was: Re: Questions about a migr=
ation to SVN & Release0.9.5)


> =20
>=20
> > -----Urspr=FCngliche Nachricht-----
> > Von: sum...@li...=20
> > [mailto:sum...@li...] Im Auftrag=20
> > von Michael Behrisch
> > Gesendet: Mittwoch, 15. November 2006 11:34
> > An: sum...@li...
> > Betreff: Re: [sumo-devel] Sorry,mistake (was: Re: Questions=20
> > about a migration to SVN & Release0.9.5)
> >=20
> > Am Mittwoch, 15. November 2006 10:33 schrieb d.k...@we...:
> > > > > > Another question that I have for a long time already=20
> > is whether=20
> > > > > > the tests shall be included in the normal SVN=20
> > repository or not.
> > > > >
> > > > > I think yes they should and they should stay where they are.
> > > > > Ideally there would be a target "tests" to make which runs all=20
> > > > > tests and fails if there are differences to the=20
> > proposed output.=20
> > > > > This would make running tests before committing easy and would=20
> > > > > help to avoid regressions.
> > > > > If TextTest is not good for that and also clumsy otherwise we=20
> > > > > should switch (go back=3F) to hand-made shell scripts.
> > >
> > > Yes, maybe we should do this. The main problem is that SUMO=20
> > contains a=20
> > > stochastic model and that there are even discrepancies between=20
> > > non-stochastic outputs when comparing versions build in=20
> > Release and in=20
> > > Debug mode using MSVC6
> >=20
> > Is this due to floating point issues or compiler optimization=3F
> > I would consider this a bug.
> >=20

Hmmm. The difference is at the fourth place after a comma, so I'm not sure=
 this is a bug. Sorry, I can not investigate this by now - time pressure.


> > > (what makes me suppose that differences will be even larger when=20
> > > comparing them to the outputs applications compiled with=20
> > gcc produce).
> >=20
> > Probably. Maybe you can give an example=3F And shouldn't the=20
> > stochastic parts behave exactly alike when provided the same=20
> > random seed=3F
> >=20

Ok, I'll try to prepare an example but not now. I'm not sure about the ran=
d() function. I suppose different compilers have different implementations=
. Sure, I though once about using a library.

> > > That's hwy I currently do not see any kind of a fast approach.=20
> > > Sure, one could evaluate each generated file, describing some=20
> > > intervals the produced value should be within but that's a=20
> > lot of work.
> >=20
> > How do you test at the moment=3F Do you look into each file and=20
> > say, well, that looks good=3F
> >=20

In principle, yes. (Sorry)
That means - I look at the files, say they're ok or not, save them in the =
test system and assure they stay the same when the code changes...

bye,
Daniel

> > > Maybe a diploma thesis should be done about it...
> >=20
> > That's always a good idea :-)
> >=20
> > > > > Does this apply to errors only or to warnings as well=3F
> > >
> > > I would like to eliminate those warnings that can be easily=20
> > > eliminated, too. As some extensions still will follow, some of the=20
> > > warnings will be removed in the future, but yes, things=20
> > like a false=20
> > > order of member-variables initialisation, comments in=20
> > comments, etc.=20
> > > should be patched.
> >=20
> > Non-virtual destructors as well, I suppose.
> >=20
> > Yours,
> > Michael
> >=20
> > --------------------------------------------------------------
> > -----------
> > Take Surveys. Earn Cash. Influence the Future of IT Join=20
> > SourceForge.net's Techsay panel and you'll get the chance to=20
> > share your opinions on IT & business topics through brief=20
> > surveys - and earn cash=20
> > http://www.techsay.com/default.php=3Fpage=3Djoin.php&p=3Dsourceforge
> &CID=3DDEVDEV
> > =5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
> > sumo-devel mailing list
> > sum...@li...
> > https://lists.sourceforge.net/lists/listinfo/sumo-devel
> >=20


=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F=5F
Der WEB.DE SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
http://smartsurfer.web.de/=3Fmc=3D100071&distributionid=3D000000000066