Re: [eigen] Static asserts in Eigen3 beta1/beta2 |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Static asserts in Eigen3 beta1/beta2
- From: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
- Date: Mon, 22 Nov 2010 08:18:07 -0500
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:received:in-reply-to :references:date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=mq5dvQvyC1fwmO5kLmk/ZltfCpC9pdHs+3hGDp877hw=; b=G9iJd/wOlOLpXJBzpJSVzuhD1wnsVGoHS1N6rAvjuDkhPUbfH2SqKsu1MMQ4rRYJAm 7HxdJxANwUiLoqMkBmPcMso98S39TmkmmN34s0mrHVH22NbWcN1QgKY7v9q+xea8BPpa QpCyu1gFR6vVLdq+XIwQirWLv1WcIqRFJMz/k=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:content-transfer-encoding; b=Ie0Ex4QEzFqYbZI95wQ495XBw+vYUIbve/+t1i+eL2PpYZx+SN5i534HGoqznz3vyg XI72eTJ7Dsnv46oFyxZoqY87GxO6Cl8xn4L1AJhqCk86Qr/vCeWDhSLTTZc9qluuqY0o s5XUbCTrH30H95XHvh4I8mkT/JIa2bT/ULkjE=
[bjacob@cahouette eigen]$ hg bisect -b
The first bad revision is:
changeset: 3382:3b87183dd7fb
user: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
date: Mon Sep 27 09:57:57 2010 -0400
summary: fix compilation on ubuntu 9.04's version of gcc 4.3 (yes, wtf)
2010/11/19 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
> This looks like a severe regression, I wasn't aware of it. Thanks for
> the report.
>
> The best next thing to do is to bisect.
>
> Benoit
>
> 2010/11/19 hauke strasdat <strasdat@xxxxxxxxx>:
>> HI Carlos,
>>
>> Thanks for your fast answer! However, I fear you got me wrong. I am
>> aware that I pass a 6-Vector into a function which expects a 5-Vector.
>> Using beta1, I get a compiler error/static assert, which is what I
>> expect and I want!
>>
>> However, the problem is - when using beta2 - I don't get a static
>> assert. The program only crashes during run-time.
>> My question is: How can I tell beta2 to trigger static asserts as beta1 did?
>>
>> Cheers,
>> Hauke
>>
>> On Fri, Nov 19, 2010 at 12:20 PM, Carlos Becker <carlosbecker@xxxxxxxxx> wrote:
>>> Hi. The problem is that your function test_v5 takes as its argument a 5x1
>>> double matrix, and you are trying to pass it a 6x1 double matrix. So the
>>> message from Eigen makes sense. If you replace Eigen::Matrix<double,5,1>
>>> by Eigen::Matrix<double,6,1> in the definition of test_v5 then it will work.
>>> The problem with this approach is that test_v5 will only work with 6x1
>>> double matrices. If you want to make it more general you can read
>>> here http://eigen.tuxfamily.org/dox-devel/TopicFunctionTakingEigenTypes.html
>>> cheers,
>>> Carlos
>>>
>>> On Fri, Nov 19, 2010 at 9:17 PM, hauke strasdat <strasdat@xxxxxxxxx> wrote:
>>>>
>>>> Hi,
>>>>
>>>> I fear this must be a known issue. However, I couldn't find any
>>>> information about it.
>>>>
>>>> Let's look at the following minimal program.
>>>>
>>>> ---
>>>> #include <Eigen/Core>
>>>>
>>>> void test_v5(const Eigen::Matrix<double,5,1> & v)
>>>> {
>>>> }
>>>>
>>>> int main()
>>>> {
>>>> Eigen::Matrix<double,6,1> v6;
>>>> test_v5(v6);
>>>> }
>>>> ---
>>>>
>>>> Compiled against Eigen3 beta1, it leads to a compiler error/static
>>>> assert: YOU_MIXED_MATRICES_OF_DIFFERENT_SIZES.
>>>> This is the desired behavior, right?
>>>>
>>>> Compiling the same program against Eigen3 beta2, however, leads to no
>>>> compiler error! Only at run-time it triggers a assertion.
>>>>
>>>> Is there any way to turn the static assertions back on in Eigen3 beta2?
>>>>
>>>> I am using Ubuntu 10.04 Amd64, with gcc 4.4.
>>>>
>>>> Thanks a lot,
>>>> Hauke
>>>>
>>>>
>>>
>>>
>>
>>
>>
>