[Bioc-devel] Generic for write to BiocGenerics

On 21 September 2017 04:15, Herv? Pag?s wrote:

If we want to move export() to BiocGenerics of course import()
would need to follow.

My rough estimate is that this move would possibly break between
20-30 packages or more. The BioC 3.6 release is coming soon and
it's a bit late to introduce such a disruptive change. So I vote
for postponing this to BioC 3.7.

@jo: Would you consider making MSnbase depend on rtracklayer in
the mean time? Alternatively, my suggestion to use a specific
write* name still stands.

writeMSData is indeed the route we decided to go in the end, so no need
to move export() and import() for us at the moment.

Thanks again for your input.

Best wishes,

Laurent

H.


On 09/18/2017 11:28 PM, Laurent Gatto wrote:

Out of curiosity, do you also have an import() function?

On 18 September 2017 22:22, Michael Lawrence wrote:

I was also about to suggest that we converge on export() but held back
because I was obviously biased. I also agree with making the target
explicit in the function name. It makes the intent of the code way
more obvious.

On Mon, Sep 18, 2017 at 3:15 PM, Herv? Pag?s <hpages at fredhutch.org> wrote:

Hi jo,

At the moment probably not much to be gained unless you ran into
some conflicts with other "write" methods defined in Bioconductor?

Note that the arguments/signature of base::write() don't really
help making it a clean generic functions (e.g. weird 'file' default,
'ncolumns' and 'sep' args that lack generality, no ellipsis). It seems
to me that instead of trying to force write() into a generic, it might
be easier/cleaner to define a method for the export() generic defined
in rtracklayer. Maybe the discussion would be whether we should consider
moving rtracklayer::export() to BiocGenerics?

Finally, IMO there is nothing wrong in using specific write* names like
writeMgfData, writeMSData, writeMzTabData, etc... It plays nicely with
tab-completion, the user can use args() to quickly see all the args and
their defaults, tab-completion also works on the arguments, and the user
does not struggle in finding the man page (like s/he does sometimes
with generic and methods, especially when those are defined in
different packages).

H.


On 09/18/2017 10:21 AM, Rainer Johannes wrote:

Dear all!

We are currently implementing mzML write support in `MSnbase` and did
implement a `write` method for the S4 objects in `MSnbase`. Now, the
question is whether it might not be better to define a `write` S4
generic in `BiocGenerics`?


cheers, jo

--
Herv? Pag?s

Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024

E-mail: hpages at fredhutch.org
Phone:  (206) 667-5791
Fax:    (206) 667-1319