[Paraview] [ParaView] Unofficial ParaView 4.1.0 SDK for Windows 32 and 64-bit (UNCLASSIFIED)

Robert Maynard robert.maynard at kitware.com
Wed Aug 13 13:28:03 EDT 2014


The fseek/ftell issue has two potential causes.

1. Somebody is using the fseek / ftell function on windows instead of
_fseeki64 / _ftelli64 which are the only way I am aware of seeking
inside files that are larger than 2GB using the C fread functions.

2. You are using std::fstream with VisualStudio 2010 or less, seekg
contains a bug where internally it casts all offsets to signed long
values, which on windows are at most 32bits. This has been fixed in VS
2012. For more information on this bug is, this SO thread is a good
starting spot: http://stackoverflow.com/questions/13837810/seekg-cannot-handle-file-of-4294967295-bytes-properly


On Wed, Aug 13, 2014 at 11:24 AM, Stephens, Michael M ERDC-RDE-ITL-MS
<Michael.M.Stephens at erdc.dren.mil> wrote:
> Classification: UNCLASSIFIED
> Caveats: FOUO
>
> some food for thought on this subject.
> VS run time system for VS2010 and earlier has a bug in it that doesn't allow XML flavored files (vtu, vtp, etc) over 4 GB (or was it 2 GB?) to be read correctly.
> it has an fseek error.  I think this was reported a while back when I first stumbled on it.
>
> I've HEARD that this was fixed in VS2012/2013 but have not verified it personally.
>
> -m
>
>
> -----Original Message-----
> From: ParaView [mailto:paraview-bounces at paraview.org] On Behalf Of Utkarsh Ayachit
> Sent: Wednesday, August 13, 2014 9:08 AM
> To: Burlen Loring
> Cc: paraview at paraview.org
> Subject: Re: [Paraview] [ParaView] Unofficial ParaView 4.1.0 SDK for Windows 32 and 64-bit
>
> Good to know. I think we'll make the switch at the same time when we upgrade the VS compiler we're using for the binaries. We are switching to using CMakified Python soon and then we wouldn't need to use VS2008. Any thoughts on which is the most common VS compiler? Qt distributes binaries for 2010, 2012, 2013. I suppose we can pick 2012 :).
>
> On Tue, Aug 12, 2014 at 11:50 PM, Burlen Loring <burlen.loring at gmail.com> wrote:
>> Ahh, that may be, I wouldn't have noticed it since I have admin on my
>> account.
>>
>> as far as the version of the standard implemented by MS MPI in HPC
>> Pack 2012 (internal ver 256), it's 2.0. The TestSystemCaps regression
>> test has the code to report.
>>
>>
>> On 8/11/2014 9:53 AM, Utkarsh Ayachit wrote:
>>>
>>> Additionally, you need to run the installer as admin since it
>>> installs a daemon (I believe). I think it would make sense to put out
>>> two version of Windows binaries (without MPI and with MS MPI) and
>>> then include the MSMPI installer as a part of the installation
>>> process, esp. since no other Windows MPI is officially supported any more.
>>>
>>> Something to think about for 4.3, not sure I want to make the plunge
>>> for 4.2 -- but we'll keep the options open :).
>>>
>>> Utkarsh
>>>
>>>
>>>
>>>
>>>
>>> On Mon, Aug 11, 2014 at 12:46 PM, Burlen Loring <bloring at lbl.gov> wrote:
>>>>
>>>> Hi Utkarsh,
>>>>
>>>> I think it does implement MPI2. At least my codes make heavy use of
>>>> MPI 2 and it works fine. However, I'll look into it and report back.
>>>>
>>>> Microsoft's documentation says:
>>>>
>>>> You may also redistribute this MS-MPI installation package with your
>>>> own applications, to facilitate having an application that can run
>>>> on a stand-alone workstation and easily take advantage of an HPC
>>>> Pack 2012 R2-based cluster. Please refer to the license agreement
>>>> for additional details.
>>>>
>>>>
>>>> Sounds like if you use an installer it could handle the dependency.
>>>>
>>>> Burlen
>>>>
>>>>
>>>> On 08/11/2014 08:03 AM, Utkarsh Ayachit wrote:
>>>>
>>>> The ParaView dev team switched from Open-MPI 1.6.2 backwards to
>>>> 1.4.4, precisely because the support was dropped.
>>>>
>>>> That is correct. That's  precisely why we've stuck with 1.4.4. I
>>>> spent a lot of time trying to packing 1.6.* on Windows to no success
>>>> and then decided to just stick with 1.4.4.
>>>>
>>>> Problem is that the source code for the current MS-MPI builds does
>>>> not seem to be publicly available,
>>>>
>>>> The problem with using MS-MPI is two fold:
>>>> 1. I can't remember if its support MPIv2 API. I don't think so, but
>>>> I may be wrong.
>>>> 2. The license (at least when I checked) doesn't allow us to packing
>>>> the dlls in our binaries. In other words, we would always have to
>>>> have the user manually install using the installer distributed by MS.
>>>>
>>>> Utkarsh
>>>>
>>>>
>>> _______________________________________________
>>> Powered by www.kitware.com
>>>
>>> Visit other Kitware open-source projects at
>>> http://www.kitware.com/opensource/opensource.html
>>>
>>> Please keep messages on-topic and check the ParaView Wiki at:
>>> http://paraview.org/Wiki/ParaView
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://public.kitware.com/mailman/listinfo/paraview
>>
>>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/paraview
>
>
>
> Classification: UNCLASSIFIED
> Caveats: FOUO
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/paraview


More information about the ParaView mailing list