Closed kbarbary closed 9 years ago
@kbarbary , thanks a lot for this PR, as I said before I find your idea of having a low-level submodule perfect. No objections about changing using FITSIO
into using FITSIO.Libcfitsio
, it's a minor change that's easy to remember.
What you propose is fine for me. I will modify OIFITS.jl
to reflect these as soon as possible. Do you plan to make a new release of FITSIO.jl
?
@emmt Yes, after this is merged would be a good time for a new release.
This PR consists of two commits:
Renames
The first commit renames some functions in the high-level API according to the style guide:
readkey
->read_key
readheader
->read_header
getcomment
->get_comment
setcomment!
->set_comment!
It's my fault that these functions weren't named this way to begin with. I was trying to mimic naming conventions in Base (e.g.,
haskey(d)
), not realizing that the style guide is a "do as we say, not as we do" sort of thing.It also deprecates the
hdu[i:j, k:l]
syntax for reading a subset of an image HDU in favor of the more explicitread(hdu, i:j, k:l)
. I really like this change as it makes all read and writes from the file explicit, making the whole API more consistent.There are deprecation warnings for all these changes.
Libcfitsio submodule
The second commit moves all of the low-level interface into a
Libcfitsio
submodule, as proposed in #27. This would require users of the low-levelfits_*
functions to dousing FITSIO.Libcfitsio
rather than justusing FITSIO
. Otherwise all low-level code would remain the same. What do users of the low-level API think about this? cc @emmt @ziotom78There are not deprecation warnings for this change, but users only need to replace
using FITSIO
withusing FITSIO.Libcfitsio
to comply with the change.