slinnarsson
commented
7 years ago I would make only two pip packages: loompy (the library) and Loom (the browser). We’re not maintaining the command-line tool, so we should remove whatever is broken and keep only the parts needed to prepare files for browsing. The command thus goes with Loom.
I can help you create the pip packages, it’s very easy.
/Sten
-- Sten Linnarsson, PhD Professor of Molecular Systems Biology Karolinska Institutet Unit of Molecular Neurobiology Department of Medical Biochemistry and Biophysics Scheeles väg 1, 171 77 Stockholm, Sweden +46 8 52 48 75 77tel://+46%208%2052%2048%2075%2077 (office) +46 70 399 32 06tel://+46%2070%20399%2032%2006 (mobile)
On 20 April 2017 at 17:46:10, Job van der Zwan (notifications@github.commailto:notifications@github.com) wrote:
So @slinnarssonhttps://github.com/slinnarsson already made a separte repository for loompy as a pure python library. This is good, but it creates the issue that the version of Loompy in this repo and that repo can go out of sync. It also is only accessible to people who install from Github.
What would make more sense (I think) is to split the code into separate units, all installable through pip, that import each other when necessary
I suggest the following split, could you comment if this is a logical separation Sten?
- loompy python library, doesn't import anything else.
- loompy-viewer constains server/browser and browser code, requires loompy (mainly for the HDF5 fallback in case expanded data is not present).
- loompy-cli installs the loom command, depends on both loompy and loompy-viewer, contains expansion code
I'm not entirely up to speed with how Python packages work, but this way (if I'm not mistaken) we can update, say loompy with something faster, and loompy-viewer and loompy-cli will just receive this update downstream after a pip install -U. No syncing inside the repositories necessary.
This should also address the issues raised in #89https://github.com/linnarsson-lab/Loom/issues/89 and #96https://github.com/linnarsson-lab/Loom/issues/96.
Sten, your thoughts on this approach?
- make loompyhttps://github.com/linnarsson-lab/loompy repository the one that is pip-installable
- make this repository pip installable as loompy-viewer, depending on loompy
- split CLI code into separate repository, installable as loompy-cli, depending on loompy and loompy-viewer
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/linnarsson-lab/Loom/issues/99, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AKKag0sDTDt_gdr26bAFHiTdeVqB8Uusks5rx33BgaJpZM4NDOB_.
JobLeonard
So @slinnarsson already made a separte repository for loompy as a pure python library. This is good, but it creates the issue that the version of Loompy in this repo and that repo can go out of sync. It also is only accessible to people who install from Github.
What would make more sense (I think) is to split the code into separate units, all installable through
pip, that import each other when necessaryI suggest the following split, could you comment if this is a logical separation Sten?
loompypython library, doesn't import anything else.loompy-viewerconstains server/browser and browser code, requires loompy (mainly for the HDF5 fallback in case expanded data is not present).loompy-cliinstalls theloomcommand, depends on bothloompyandloompy-viewer, contains expansion codeI'm not entirely up to speed with how Python packages work, but this way (if I'm not mistaken) we can update, say
loompywith something faster, andloompy-viewerandloompy-cliwill just receive this update downstream after apip install -U. No syncing inside the repositories necessary.This should also address the issues raised in #89 and #96.
Sten, your thoughts on this approach?
loompyrepository the one that is pip-installableloompy-viewer, depending onloompysplit CLI code into separate repository, installable asloompy-cli, depending onloompyandloompy-viewerloom-viewer