Sorry, I don't like the abbreviation, I love change :-)<br><br><br><div class="gmail_quote">On Thu, Jan 28, 2010 at 8:45 AM, Jeff Baumes <span dir="ltr"><<a href="mailto:jeff.baumes@kitware.com">jeff.baumes@kitware.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="im">On Wed, Jan 27, 2010 at 10:58 AM, Will Schroeder<br>
<<a href="mailto:will.schroeder@kitware.com">will.schroeder@kitware.com</a>> wrote:<br>
> VTK has a verbose, self documenting style (for better or worse). I'd like to<br>
> stick with it if possible.<br>
<br>
</div>Will,<br>
<br>
Are you suggesting no change? Or just that you don't like the<br>
abbreviated typedefs like vtkRendererSP?<br>
<br>
Jeff<br>
<div><div></div><div class="h5"><br>
><br>
><br>
> On Wed, Jan 27, 2010 at 10:56 AM, David Cole <<a href="mailto:david.cole@kitware.com">david.cole@kitware.com</a>> wrote:<br>
>><br>
>> On Wed, Jan 27, 2010 at 8:35 AM, Jeff Baumes <<a href="mailto:jeff.baumes@kitware.com">jeff.baumes@kitware.com</a>><br>
>> wrote:<br>
>>><br>
>>> On Tue, Jan 26, 2010 at 12:44 PM, Marcus D. Hanwell<br>
>>> <<a href="mailto:marcus.hanwell@kitware.com">marcus.hanwell@kitware.com</a>> wrote:<br>
>>> > On Friday 08 January 2010 16:46:47 David Cole wrote:<br>
>>> >> On Fri, Jan 8, 2010 at 4:08 PM, Marcus D. Hanwell <<br>
>>> >><br>
>>> >> <a href="mailto:marcus.hanwell@kitware.com">marcus.hanwell@kitware.com</a>> wrote:<br>
>>> >> > I would like to add my -1, I think needing to split lines in order<br>
>>> >> > to<br>
>>> >> > declare<br>
>>> >> > a new local variable is a little much. I came from a C++ background<br>
>>> >> > where<br>
>>> >> > any<br>
>>> >> > object could be declared on the stack though. For things like the<br>
>>> >> > examples it<br>
>>> >> > seems to hurt readability to me.<br>
>>> >> ><br>
>>> >> > Pointer:<br>
>>> >> > vtkFloatArray *myTable = vtkFloatArray::New();<br>
>>> >> > myTable->Delete();<br>
>>> >> > myTable = NULL;<br>
>>> >> ><br>
>>> >> > Smart pointer:<br>
>>> >> > vtkSmartPointer<vtkFloatArray> myTable =<br>
>>> >> > vtkSmartPointer<vtkFloatArray>::New();<br>
>>> >> ><br>
>>> >> > Smart pointer with macro:<br>
>>> >> > VTK_CREATE(vtkFloatArray, myTable);<br>
>>> >> ><br>
>>> >> > Stack:<br>
>>> >> > vtkFloatArray myTable;<br>
>>> >> ><br>
>>> >> > I would prefer to be able to use something like the first or the<br>
>>> >> > last. In<br>
>>> >> > classes etc it is often a different story. It seems like there<br>
>>> >> > should be<br>
>>> >> > some<br>
>>> >> > macro or template function to generate variables with less<br>
>>> >> > repetition.<br>
>>> >><br>
>>> >> Prefer the first and last as much as you want, but we simply can't use<br>
>>> >> them<br>
>>> >> in VTK. The first leads to memory leaks because people forget the<br>
>>> >> Delete<br>
>>> >> calls. The last cannot be done with vtkObject derived classes because<br>
>>> >> of<br>
>>> >> the nature of vtkObject reference counting...<br>
>>> >><br>
>>> > Wasn't suggesting either (just pointing out the shorter syntax that<br>
>>> > people<br>
>>> > miss), although the first is still widely used in VTK.<br>
>>> ><br>
>>> >> So we have to pick one of the middle ones...<br>
>>> >><br>
>>> >> It's unfortunate that we've had two +1's and two -1's... that leaves<br>
>>> >> us at<br>
>>> >> 0 for the moment. I guess that and the fact that it's Friday makes it<br>
>>> >> fairly easy to put off a decision until at least next week. ;-)<br>
>>> >><br>
>>> >> *If* we do go with a macro-based approach, I think we can all agree<br>
>>> >> there<br>
>>> >> should be one centralized macro that does this and it should be used<br>
>>> >> *everywhere* that vtkSmartPointer::New is presently used.<br>
>>> ><br>
>>> > What about a vtkLocalPointer<vtkClass> myLocal; where the default<br>
>>> > constructor<br>
>>> > makes an instance on the VTK class? It would also be possible to have a<br>
>>> > constructor take an argument (may be a little clunkier) such as<br>
>>> > vtkSmartPointer<vtkClass> myLocal(true); if we do not want to introduce<br>
>>> > yet<br>
>>> > another class.<br>
>>> ><br>
>>> > Alternate options a and b...<br>
>>> ><br>
>>> > a) vtkLocalPointer<vtkClass> myLocal;<br>
>>> > b) vtkSmartPointer<vtkClass> myLocal(true);<br>
>>> ><br>
>>> > Would this be preferable to a macro? It seems like a better way to go<br>
>>> > to me,<br>
>>> > and in terms of API and conciseness seems to satisfy our requirements<br>
>>> > better<br>
>>> > than the current approach. It would still be possible to share the<br>
>>> > pointer<br>
>>> > too, due to the reference counting in vtkObject derived classes.<br>
>>><br>
>>> I think option b is a good option. It seems that it would be odd to<br>
>>> have two different pointer classes, where the only difference is what<br>
>>> they do on construction.<br>
>>><br>
>>> The biggest grudge I have about the current option is that it almost<br>
>>> always requires a split line if you constrain yourself to 80<br>
>>> characters. If your type name is more than about 15 characters (which<br>
>>> many VTK types are), you have to split your line:<br>
>>><br>
>>> vtkSmartPointer<type> name = vtkSmartPointer<type>::New();<br>
>>> 16 + t + 2 + n + 19 + t + 9 = 46 + 2*t + n<br>
>>><br>
>>> Option b, even though not the most concise, would make it easy to stay<br>
>>> within that limit, about halving the number of characters:<br>
>>><br>
>>> vtkSmartPointer<type> name(true);<br>
>>> 16 + t + 2 + n + 7 = 25 + t + n<br>
>>><br>
>>> Jeff<br>
>><br>
>><br>
>> Another solution worth considering is simply making a typedef called<br>
>> "vtkRendererSP" that is a typedef for "vtkSmartPointer<vtkRenderer>" -- that<br>
>> would allow you to:<br>
>> vtkRendererSP renderer = vtkRendererSP::New();<br>
>> It gives you the shorter names you're all longing for, without changing<br>
>> anything else already in VTK...<br>
>><br>
>> David C.<br>
>><br>
>> _______________________________________________<br>
>> Powered by <a href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
>><br>
>> Visit other Kitware open-source projects at<br>
>> <a href="http://www.kitware.com/opensource/opensource.html" target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
>><br>
>> Follow this link to subscribe/unsubscribe:<br>
>> <a href="http://www.vtk.org/mailman/listinfo/vtk-developers" target="_blank">http://www.vtk.org/mailman/listinfo/vtk-developers</a><br>
>><br>
>><br>
><br>
><br>
><br>
> --<br>
> William J. Schroeder, PhD<br>
> Kitware, Inc.<br>
> 28 Corporate Drive<br>
> Clifton Park, NY 12065<br>
> <a href="mailto:will.schroeder@kitware.com">will.schroeder@kitware.com</a><br>
> <a href="http://www.kitware.com" target="_blank">http://www.kitware.com</a><br>
> (518) 881-4902<br>
><br>
<br>
<br>
<br>
</div></div>--<br>
Jeff Baumes, Ph.D.<br>
<div class="im">R&D Engineer, Kitware Inc.<br>
</div>(518) 881-4932<br>
<font color="#888888"><a href="mailto:jeff.baumes@kitware.com">jeff.baumes@kitware.com</a><br>
</font></blockquote></div><br><br clear="all"><br>-- <br>William J. Schroeder, PhD<br>Kitware, Inc.<br>28 Corporate Drive<br>Clifton Park, NY 12065<br><a href="mailto:will.schroeder@kitware.com">will.schroeder@kitware.com</a><br>
<a href="http://www.kitware.com">http://www.kitware.com</a><br>(518) 881-4902<br>
<input id="gwProxy" type="hidden"><input onclick="jsCall();" id="jsProxy" type="hidden"><div id="refHTML"></div>