Re: [Seqan-dev] placing seqan headers into R package

[Rahn, René <Rene.Rahn@fu-berlin.de>]

Hi August,

many thanks for your efforts. I am afraid, that this is actually quite a challenge you are trying here.

We once had support for MinGW, but it caused more problems than bringing any benefit. Hence we removed the support from it, also given by the fact that there was always a free version of Visual Studio available.

Of course, if you are eager to tackle this you will get support from us. But honestly, we need to really check if we can add this into the core library if it will even work.

Please note, that there will be only patch level releases left for the 2.x cycle. So having this MinGW port in a separate repository could be a convenient way for both sides.

We are currently working on SeqAn 3 and this will be no compatible with SeqAn 2 and requires c++17/20. I am afraid this will be a stopper for MinGW for quite some time.

But, I think if Windows comes with a linux subsystem these days it could be worth it to check if R might  package using standard linux on Windows anyway in the future (Not sure if they have it on their wishlist already).

However, we are still happy to assist you to try out minGW compiler. Please let us know the compiler errors you receive, and the invoked build command, so we can have a closer look on our side.

For the respective error you are receiving, it seems the only place that is using it is the <seqan/system/file_sync.h> header and some other headers doing some forward declaration of it.

As far as I am concerned, this is old code and does not really need to be shipped in your project. So you could wrap it into macros, that if MinGW is used these code paths are not evaluated.

We have a virtual stream class now, which is a better alternative for the File class.

Anyhow, I couldn't find any place where we directly use File<Sync|Async> for (a)synchronous I/O so it would be safe to remove for the R port.

Let me know if there is anything more, but please bear in mind, that we are quite busy right now, and it could take a while for us to respond with a proper solution.

Especially in the next two weeks ;)

Best,

René

On 5. Mar 2018, at 18:01, August Guang <august.guang@gmail.com> wrote:

Thanks, I have included the license file and a notice about citations now. Will there be support for seqan/2.4 for mingw? I saw that there was on older versions. The reason I'm asking is that R packages submitted to Bioconductor/CRAN have to build and pass checks on Linux, OS and Windows, but on Windows the packages are tested with the mingw compiler and I haven't found a way around it.

If not, would I be able to get some help with writing support for it? I have been trying to address various issues specific to mingw but am not familiar with Windows specific implementations and am getting way out of my depth. So far I have added mman-win32 to replace sys/mman.h and changed some enum values to not use posix_madvise. The current error I am trying to tackle is the fact that getuid, geteuid, getgid and getegid are Unix-only. It appears I can either write my own macros for them (https://stackoverflow.com/questions/1594746/win32-equivalent-of-getuid) or install LibGW32 for them (http://gnuwin32.sourceforge.net/packages/libgw32c.htm). Is there an option that would be better and can I get some advice on how I would implement it?

thanks,

august

On Mon, Feb 26, 2018 at 7:49 AM, Hannes Hauswedell <hannes.hauswedell@fu-berlin.de> wrote:

Am Donnerstag, 22. Februar 2018, 11:05:38 schrieb August Guang:
> Hi,
>
> I would like to package the SeqAn headers for easy use in a few R packages
> I will be writing with my team, the Computational Biology Core at Brown
> University in the US. Would this be ok with you all? I am leaving the
> license as BSD 3 clause as well, and if you would like to make any
> additions to the license please let me know.
>
> You can view the current state of the repository here:
> https://github.com/compbiocore/RSeqAn
>
> Thanks,
> august

Dear August,

that's great to hear, feel very free to include it in your package. As long as
the LICENSE file is shipped with the package there is no problem. Please also
include a link to www.seqan.de in your documentation and if possible also the
following note:
"Please do not forget to cite SeqAn if you use this package in your acedemic
work, even if only in auxiliary tools or pipelines:
https://doi.org/10.1016/j.jbiotec.2017.07.017"

That would be great!

Thank you,
Hannes
--
Hannes Hauswedell

Scientific staff & PhD candidate
Freie Universität Berlin / Max Planck Institute for Molecular Genetics

address     Institut für Informatik
            Takustraße 9
            Room 019
            14195 Berlin
telephone   +49 (0)30 838-75241
fax         +49 (0)30 838-75218
e-mail      hannes.hauswedell@fu-berlin.de

-- 

August Guang

HMC Class of 2012

Mathematics

Brown University

Applied Mathematics

_______________________________________________
seqan-dev mailing list
seqan-dev@lists.fu-berlin.de
https://lists.fu-berlin.de/listinfo/seqan-dev

---

René Rahn

Ph.D. Student (de.NBI - CIBI)

--------------------------------

Tel:  (+49) 30 838 72974

Mail: rene.rahn@fu-berlin.de

--------------------------------

Institute of Computer Science

Algorithmic Bioinformatics (ABI)

--------------------------------

Freie Universität Berlin

Takustraße 9

14195 Berlin

--------------------------------