| Re: [eigen] Question about ColMajor vs. RowMajor |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Question about ColMajor vs. RowMajor
- From: Rhys Ulerich <rhys.ulerich@xxxxxxxxx>
- Date: Mon, 19 Dec 2011 09:27:07 -0600
- Cc: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>, Michael Dixon <mdixon@xxxxxxxxxxxxxxxx>, Suat Gedikli <gedikli@xxxxxxxxxxxxxxxx>
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=WsqOUeIrP2ZCsfEQfVYW9bn5VEqypo1tOz/AH29ZisM=; b=J0IXGb0xa8ZWKUz+AUTkoMaMdPVgbaZTjJGzuPae6Yi+eJpymtvaKJFIY6f1FLw1QR 8xVLvbTT7n3Vd51oOlF9vrZ5nGWXiUa2nvc5rPHG1/MumThQRvqgsCMfgGuWsKtMmaey 80T/xxVWcaRYaRMbyPm90MOrCkrBhn5wkFbdc=
> For example, if people misuse the () operator and are not reading
> the documentation properly, then every time someone upstream
> adds the -DEIGEN_DEFAULT_TO_ROW_MAJOR flag, their code
> will break in horrible ways without a single warning.
>
> ...
>
> I am still a bit conflicted on what to tell users.
If the storage ordering of the matrix matters for one of your users'
uses of operator(), add some compile-time assertions to check that the
incoming types are indeed row or column major. Those regions should
have such asserts anyhow to make the storage requirements
self-documenting.
There's no fairy dust that can protect people from these types of mistakes.
- Rhys