Reduce size of pip'd distribution

[delucchi-cmu]

Might include tests and test data? Or this is just the size of the dust maps?

[hombit]

Dust maps idea sounds right to me:

pip download superphot-plus
unzip superphot_plus-0.0.2-py3-none-any.whl /tmp/superphot-plus
find /tmp/superphot-plus -type f | xargs ls -lk

-rw-r--r--  1 hombit  wheel      1071 Aug 10 19:19 /tmp/superphot-plus/superphot_plus-0.0.2.dist-info/LICENSE
-rw-r--r--  1 hombit  wheel      5635 Aug 10 19:19 /tmp/superphot-plus/superphot_plus-0.0.2.dist-info/METADATA
-rw-rw-r--  1 hombit  wheel      2963 Aug 10 19:20 /tmp/superphot-plus/superphot_plus-0.0.2.dist-info/RECORD
-rw-r--r--  1 hombit  wheel        92 Aug 10 19:19 /tmp/superphot-plus/superphot_plus-0.0.2.dist-info/WHEEL
-rw-r--r--  1 hombit  wheel        15 Aug 10 19:19 /tmp/superphot-plus/superphot_plus-0.0.2.dist-info/top_level.txt
-rw-r--r--  1 hombit  wheel        30 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/__init__.py
-rw-r--r--  1 hombit  wheel       160 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/_version.py
-rw-r--r--  1 hombit  wheel     14925 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/classify_ztf.py
-rw-r--r--  1 hombit  wheel      1226 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/constants.py
-rw-r--r--  1 hombit  wheel         0 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/data_generation/__init__.py
-rw-r--r--  1 hombit  wheel      5433 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/data_generation/alerce.py
-rw-r--r--  1 hombit  wheel      3630 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/data_generation/antares.py
-rw-r--r--  1 hombit  wheel         0 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/data_generation/elasticc.py
-rw-r--r--  1 hombit  wheel      7968 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/data_generation/make_fake_spp_data.py
-rw-r--r--  1 hombit  wheel       856 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/file_paths.py
-rw-r--r--  1 hombit  wheel      2900 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/file_utils.py
-rw-r--r--  1 hombit  wheel      5211 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/format_data_ztf.py
-rw-r--r--  1 hombit  wheel      5666 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/import_utils.py
-rw-r--r--  1 hombit  wheel     11214 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/lightcurve.py
-rw-r--r--  1 hombit  wheel     14715 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/mlp.py
-rw-r--r--  1 hombit  wheel     34610 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/plotting.py
-rw-r--r--  1 hombit  wheel      2316 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/posterior_samples.py
-rw-r--r--  1 hombit  wheel      6585 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/samplers/dynesty_sampler.py
-rw-r--r--  1 hombit  wheel     11415 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/samplers/numpyro_sampler.py
-rw-r--r--  1 hombit  wheel       500 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/samplers/sampler.py
-rw-r--r--  1 hombit  wheel  67115520 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/sfd/SFD_dust_4096_ngp.fits
-rw-r--r--  1 hombit  wheel  67115520 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/sfd/SFD_dust_4096_sgp.fits
-rw-r--r--  1 hombit  wheel       184 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/sfd/__init__.py
-rw-r--r--  1 hombit  wheel      5976 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/supernova_class.py
-rw-r--r--  1 hombit  wheel      3089 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/surveys/fitting_priors.py
-rw-r--r--  1 hombit  wheel      3171 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/surveys/surveys.py
-rw-r--r--  1 hombit  wheel      1454 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/surveys/ztf.yaml
-rw-r--r--  1 hombit  wheel      7057 Aug 10 19:19 /tmp/superphot-plus/superphot_plus/utils.py

[delucchi-cmu]

@camposandro pointed out that there's a limit on pypi of the distribution size, and we're pretty close to it (something like 100M, and superphot-plus is 90M).

[camposandro]

@delucchi-cmu Unfortunately I haven't been able to find any official remark regarding the size limit. It looks like the default used to be 60MB (after compression), but we are already exceeding that amount by a lot!

@kdesoto-astro Anyway, would it be feasible to delete the dust maps and replace them with a query to their remote service? To be honest this would raise other questions, I'm not sure it's a good idea to rely on the availability of this external service.

[hombit]

@camposandro from other project