[ITP] python-h5py 2.2.0

[ITP] python-h5py 2.2.0

[Chris LeBlanc]

Hello,

H5py (www.h5py.org) is a useful package for managing large scientific
datasets in Python.  The hdf5 library packages already exist in
Cygwin, so there are no additional dependencies and no changes were
needed to compile the C Python extension modules under Cygwin.
Maintaining this package should be pretty light work and I'm happy to
do it.

H5py is a standard package in Debian and Fedora:
http://packages.debian.org/sid/python-h5py
https://admin.fedoraproject.org/pkgdb/acls/name/h5py

I used Cygport to create the archives and setup.hint for version 2.2.0
of h5py.  This is the first time I've used cygport so please let me
know if I've done something incorrectly.  I haven't seen a single repo
for keeping track of these pacakge config files, so made my own github
repo for this file: https://github.com/crleblanc/cygwin_h5py_package.

The h5py package includes a set of unit tests (run with
"h5py.run_tests()" in a Python shell).  I've run these after building
and installing, and they run fine on both 32 and 64 bit.  I haven't
run any large scale tests.

Here are the packages I've created on fresh 32 and 64 bit Cygwin
instances.  I see there is a Cygwin web page for uploading packages,
but this package hasn't been accepted yet so thought I'd do it the old
fashioned way.

32bit:
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-2.2.0-1-src.tar.xz
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-2.2.0-1.tar.xz
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/setup.hint
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-debuginfo/python-h5py-debuginfo-2.2.0-1.tar.xz
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-debuginfo/setup.hint

64bit:
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-2.2.0-1-src.tar.xz
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-2.2.0-1.tar.xz
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/setup.hint
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-debuginfo/python-h5py-debuginfo-2.2.0-1.tar.xz
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-debuginfo/setup.hint

Setup.hint:

category: Python
requires: libhdf5_8 python python-numpy hdf5
sdesc: "A Pythonic interface to the HDF5 binary data format"
ldesc: "This package lets you store huge amounts of numerical
data, and easily manipulate that data from NumPy. For example, you
can slice into multi-terabyte datasets stored on disk, as if they
were real NumPy arrays. Thousands of datasets can be stored in a
single file, categorized and tagged however you want."

Hopefully this is all the info you need.  I noticed a possible
dependency issue with NumPy, but I'll post that as a different message
on the appropriate list.

Regards,
Chris LeBlanc
GNS Science

Re: [ITP] python-h5py 2.2.0

[Chris LeBlanc]

It's been several weeks since I posted this ITP and there hasn't been
a response.

Version 2.2.1 of h5py is out, and I'm willing to package it for Cygwin
but I'll need to know if this package will be accepted.  Please let me
know if I'm doing something incorrectly or if I should make any
changes.

I know there is a new package upload system, but it appears to be for
accepted packages only.  Please let me know if I can use this upload
system for this proposed package.

Cheers,
Chris

On Wed, Dec 11, 2013 at 12:03 PM, Chris LeBlanc <crleblanc@gmail.com> wrote:
> Hello,
>
> H5py (www.h5py.org) is a useful package for managing large scientific
> datasets in Python.  The hdf5 library packages already exist in
> Cygwin, so there are no additional dependencies and no changes were
> needed to compile the C Python extension modules under Cygwin.
> Maintaining this package should be pretty light work and I'm happy to
> do it.
>
> H5py is a standard package in Debian and Fedora:
> http://packages.debian.org/sid/python-h5py
> https://admin.fedoraproject.org/pkgdb/acls/name/h5py
>
> I used Cygport to create the archives and setup.hint for version 2.2.0
> of h5py.  This is the first time I've used cygport so please let me
> know if I've done something incorrectly.  I haven't seen a single repo
> for keeping track of these pacakge config files, so made my own github
> repo for this file: https://github.com/crleblanc/cygwin_h5py_package.
>
> The h5py package includes a set of unit tests (run with
> "h5py.run_tests()" in a Python shell).  I've run these after building
> and installing, and they run fine on both 32 and 64 bit.  I haven't
> run any large scale tests.
>
> Here are the packages I've created on fresh 32 and 64 bit Cygwin
> instances.  I see there is a Cygwin web page for uploading packages,
> but this package hasn't been accepted yet so thought I'd do it the old
> fashioned way.
>
> 32bit:
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-2.2.0-1-src.tar.xz
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-2.2.0-1.tar.xz
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/setup.hint
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-debuginfo/python-h5py-debuginfo-2.2.0-1.tar.xz
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/32bit/python-h5py/python-h5py-debuginfo/setup.hint
>
> 64bit:
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-2.2.0-1-src.tar.xz
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-2.2.0-1.tar.xz
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/setup.hint
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-debuginfo/python-h5py-debuginfo-2.2.0-1.tar.xz
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/64bit/python-h5py/python-h5py-debuginfo/setup.hint
>
> Setup.hint:
>
> category: Python
> requires: libhdf5_8 python python-numpy hdf5
> sdesc: "A Pythonic interface to the HDF5 binary data format"
> ldesc: "This package lets you store huge amounts of numerical
> data, and easily manipulate that data from NumPy. For example, you
> can slice into multi-terabyte datasets stored on disk, as if they
> were real NumPy arrays. Thousands of datasets can be stored in a
> single file, categorized and tagged however you want."
>
> Hopefully this is all the info you need.  I noticed a possible
> dependency issue with NumPy, but I'll post that as a different message
> on the appropriate list.
>
> Regards,
> Chris LeBlanc
> GNS Science

Re: [ITP] python-h5py 2.2.0

[Marco Atzeri]

On 15/01/2014 03:57, Chris LeBlanc wrote:
> It's been several weeks since I posted this ITP and there hasn't been
> a response.
>
> Version 2.2.1 of h5py is out, and I'm willing to package it for Cygwin
> but I'll need to know if this package will be accepted.  Please let me
> know if I'm doing something incorrectly or if I should make any
> changes.
>
> I know there is a new package upload system, but it appears to be for
> accepted packages only.  Please let me know if I can use this upload
> system for this proposed package.
>
> Cheers,
> Chris

Hi Chris,
sorry, I missed your previous request.

To download , better in this way

wget -r -np -nH --cut-dirs=4 
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/

so we can copy and download in one shot.

It builds and package fine on 32bit and 64bit,
but there is no test check.
I assume you already tested it for your own correct ?

If so GTG for me. As it is already in Debian and Fedora we need no vote.

Questions:
- as 2.2.1 is out will you repack ?
- are you planning to make also python3 version ?



>>
>> Setup.hint:
>>
>> category: Python
>> requires: libhdf5_8 python python-numpy hdf5
>> sdesc: "A Pythonic interface to the HDF5 binary data format"
>> ldesc: "This package lets you store huge amounts of numerical
>> data, and easily manipulate that data from NumPy. For example, you
>> can slice into multi-terabyte datasets stored on disk, as if they
>> were real NumPy arrays. Thousands of datasets can be stored in a
>> single file, categorized and tagged however you want."
>>
>> Hopefully this is all the info you need.  I noticed a possible
>> dependency issue with NumPy, but I'll post that as a different message
>> on the appropriate list.
>>
>> Regards,
>> Chris LeBlanc
>> GNS Science

Regards
Marco

PS: nice to see that my hdf5 package is useful outside octave

Re: [ITP] python-h5py 2.2.0

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 914 bytes --]

On Jan 15 07:25, Marco Atzeri wrote:
> On 15/01/2014 03:57, Chris LeBlanc wrote:
> >It's been several weeks since I posted this ITP and there hasn't been
> >a response.
> >
> >Version 2.2.1 of h5py is out, and I'm willing to package it for Cygwin
> >but I'll need to know if this package will be accepted.  Please let me
> >know if I'm doing something incorrectly or if I should make any
> >changes.
> >
> >I know there is a new package upload system, but it appears to be for
> >accepted packages only.  Please let me know if I can use this upload
> >system for this proposed package.

You can.  Just have a read at 
https://sourceware.org/cygwin-apps/package-upload.html
and send the public part of an ssh key as outlined.


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 819 bytes --]

Re: [ITP] python-h5py 2.2.0

[Christopher Faylor]

On Wed, Jan 15, 2014 at 10:36:06AM +0100, Corinna Vinschen wrote:
>On Jan 15 07:25, Marco Atzeri wrote:
>> On 15/01/2014 03:57, Chris LeBlanc wrote:
>> >It's been several weeks since I posted this ITP and there hasn't been
>> >a response.
>> >
>> >Version 2.2.1 of h5py is out, and I'm willing to package it for Cygwin
>> >but I'll need to know if this package will be accepted.  Please let me
>> >know if I'm doing something incorrectly or if I should make any
>> >changes.
>> >
>> >I know there is a new package upload system, but it appears to be for
>> >accepted packages only.  Please let me know if I can use this upload
>> >system for this proposed package.
>
>You can.  Just have a read at 
>https://sourceware.org/cygwin-apps/package-upload.html
>and send the public part of an ssh key as outlined.

Actually, you can't use the upload system to upload packages that aren't
accepted.  This is only for packages which are listed in
cygwin-pkg-maint.  So, yes, the upload system is only intended for
packages that are accepted.  It isn't intended to be used as temporary
staging for packages that might be accepted.

Re: [ITP] python-h5py 2.2.0

[Chris LeBlanc]

Hi Marco,

Thanks for your feedback.  I've rebuilt the packages for 2.2.1, and
also for Python 3.

I was running the unit test suite manually after installing h5py by
calling h5py.run_tests().  I've added this to the cygport files under
the src_test() function so we can now test using cygport.  I had to
run the unit tests under the staging directory instead of the build
directory, because Cython builds the extension module .dll files in a
temporary build directory, therefore we can't import h5py from the
build dir.  The tests pass for Python 2 and 3 for both 32 and 64 bit.

I've updated the files on the FTP site.  Thanks for the one-line wget
command, the same command can be used to download the updated files:
wget -r -np -nH --cut-dirs=4
ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/

Thanks for working on the hdf5 package, I was pleased to see this in
Cygwin.  I think it could be useful for quite a few packages.

Here is the setup.hint for the Python 3 version of h5py.  The only
change here is switching from python/python-numpy to
python3/python3-numpy:

category: Python
requires: libhdf5_8 python3 python3-numpy hdf5
sdesc: "A Pythonic interface to the HDF5 binary data format for Python 3"
ldesc: "This package lets you store huge amounts of numerical
data, and easily manipulate that data from NumPy. For example, you
can slice into multi-terabyte datasets stored on disk, as if they
were real NumPy arrays. Thousands of datasets can be stored in a
single file, categorized and tagged however you want."


Hopefully this is now ready to go.  Please let me know if there's anything else.

Thanks again,
Chris

On Wed, Jan 15, 2014 at 7:25 PM, Marco Atzeri <marco.atzeri@gmail.com> wrote:
> On 15/01/2014 03:57, Chris LeBlanc wrote:
>>
>> It's been several weeks since I posted this ITP and there hasn't been
>> a response.
>>
>> Version 2.2.1 of h5py is out, and I'm willing to package it for Cygwin
>> but I'll need to know if this package will be accepted.  Please let me
>> know if I'm doing something incorrectly or if I should make any
>> changes.
>>
>> I know there is a new package upload system, but it appears to be for
>> accepted packages only.  Please let me know if I can use this upload
>> system for this proposed package.
>>
>> Cheers,
>> Chris
>
>
> Hi Chris,
> sorry, I missed your previous request.
>
> To download , better in this way
>
> wget -r -np -nH --cut-dirs=4
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/
>
> so we can copy and download in one shot.
>
> It builds and package fine on 32bit and 64bit,
> but there is no test check.
> I assume you already tested it for your own correct ?
>
> If so GTG for me. As it is already in Debian and Fedora we need no vote.
>
> Questions:
> - as 2.2.1 is out will you repack ?
> - are you planning to make also python3 version ?
>
>
>
>
>>>
>>> Setup.hint:
>>>
>>> category: Python
>>> requires: libhdf5_8 python python-numpy hdf5
>>> sdesc: "A Pythonic interface to the HDF5 binary data format"
>>> ldesc: "This package lets you store huge amounts of numerical
>>> data, and easily manipulate that data from NumPy. For example, you
>>> can slice into multi-terabyte datasets stored on disk, as if they
>>> were real NumPy arrays. Thousands of datasets can be stored in a
>>> single file, categorized and tagged however you want."
>>>
>>> Hopefully this is all the info you need.  I noticed a possible
>>> dependency issue with NumPy, but I'll post that as a different message
>>> on the appropriate list.
>>>
>>> Regards,
>>> Chris LeBlanc
>>> GNS Science
>
>
> Regards
> Marco
>
> PS: nice to see that my hdf5 package is useful outside octave

Re: [ITP] python-h5py 2.2.0

[Marco Atzeri]

On 16/01/2014 03:36, Chris LeBlanc wrote:
> Hi Marco,
>
> Thanks for your feedback.  I've rebuilt the packages for 2.2.1, and
> also for Python 3.
>
> I was running the unit test suite manually after installing h5py by
> calling h5py.run_tests().  I've added this to the cygport files under
> the src_test() function so we can now test using cygport.  I had to
> run the unit tests under the staging directory instead of the build
> directory, because Cython builds the extension module .dll files in a
> temporary build directory, therefore we can't import h5py from the
> build dir.  The tests pass for Python 2 and 3 for both 32 and 64 bit.
>
> I've updated the files on the FTP site.  Thanks for the one-line wget
> command, the same command can be used to download the updated files:
> wget -r -np -nH --cut-dirs=4
> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/
>
> Thanks for working on the hdf5 package, I was pleased to see this in
> Cygwin.  I think it could be useful for quite a few packages.
>
> Here is the setup.hint for the Python 3 version of h5py.  The only
> change here is switching from python/python-numpy to
> python3/python3-numpy:
>
> category: Python
> requires: libhdf5_8 python3 python3-numpy hdf5
> sdesc: "A Pythonic interface to the HDF5 binary data format for Python 3"
> ldesc: "This package lets you store huge amounts of numerical
> data, and easily manipulate that data from NumPy. For example, you
> can slice into multi-terabyte datasets stored on disk, as if they
> were real NumPy arrays. Thousands of datasets can be stored in a
> single file, categorized and tagged however you want."
>
>
> Hopefully this is now ready to go.  Please let me know if there's anything else.
>
> Thanks again,
> Chris


for me it was already GTG (Good TO Go) and I had the impression also
Corinna had no objection.

Justed rebuilt your python-h5py-2.2.1-1 and it builds and packages fine.

please follow
https://sourceware.org/cygwin-apps/package-upload.html

for upload instruction
Regards
Marco

Re: [ITP] python-h5py 2.2.0

[Chris LeBlanc]

Hi Marco,

That's excellent, thanks for having another look.  The newer
instructions you linked to are fine, but the old instructions for
submitting a package don't refer to the new upload method
(http://cygwin.com/setup.html).  Not a big deal, but may cause some
confusion for other people adding new packages.

I'll mail my SSH public key to this list as per the instructions,
upload the packages once I get confirmation of my key, and send an
announcement to cygwin-announce once that's finished.  Hopefully
that's it, but please let me know if there's anything else I should
do.

Thanks again,
Chris

On Fri, Jan 24, 2014 at 6:51 PM, Marco Atzeri <marco.atzeri@gmail.com> wrote:
>
>
> On 16/01/2014 03:36, Chris LeBlanc wrote:
>>
>> Hi Marco,
>>
>> Thanks for your feedback.  I've rebuilt the packages for 2.2.1, and
>> also for Python 3.
>>
>> I was running the unit test suite manually after installing h5py by
>> calling h5py.run_tests().  I've added this to the cygport files under
>> the src_test() function so we can now test using cygport.  I had to
>> run the unit tests under the staging directory instead of the build
>> directory, because Cython builds the extension module .dll files in a
>> temporary build directory, therefore we can't import h5py from the
>> build dir.  The tests pass for Python 2 and 3 for both 32 and 64 bit.
>>
>> I've updated the files on the FTP site.  Thanks for the one-line wget
>> command, the same command can be used to download the updated files:
>> wget -r -np -nH --cut-dirs=4
>> ftp://ftp.gns.cri.nz/pub/claritas/misc/cygwin_h5py_packages/
>>
>> Thanks for working on the hdf5 package, I was pleased to see this in
>> Cygwin.  I think it could be useful for quite a few packages.
>>
>> Here is the setup.hint for the Python 3 version of h5py.  The only
>> change here is switching from python/python-numpy to
>> python3/python3-numpy:
>>
>> category: Python
>> requires: libhdf5_8 python3 python3-numpy hdf5
>> sdesc: "A Pythonic interface to the HDF5 binary data format for Python 3"
>> ldesc: "This package lets you store huge amounts of numerical
>> data, and easily manipulate that data from NumPy. For example, you
>> can slice into multi-terabyte datasets stored on disk, as if they
>> were real NumPy arrays. Thousands of datasets can be stored in a
>> single file, categorized and tagged however you want."
>>
>>
>> Hopefully this is now ready to go.  Please let me know if there's anything
>> else.
>>
>> Thanks again,
>> Chris
>
>
>
> for me it was already GTG (Good TO Go) and I had the impression also
> Corinna had no objection.
>
> Justed rebuilt your python-h5py-2.2.1-1 and it builds and packages fine.
>
> please follow
> https://sourceware.org/cygwin-apps/package-upload.html
>
> for upload instruction
> Regards
> Marco
>