Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

flatpak: use cached versions from pypi instead of git #106

Closed
wants to merge 1 commit into from

Conversation

bentiss
Copy link
Collaborator

@bentiss bentiss commented Mar 6, 2018

Each time we redo the flatpak, we pull the files from git, which is less than optimal.

Rely on the cache provided by pypi to have consistent files.

Not sure if this is allowed or reliable, so asking everyone if this is acceptable.

Each time we redo the flatpak, we pull the files from git, which is less
than optimal.

Rely on the cache provided by pypi to have consistent files.
@bentiss bentiss added the question Further information is requested label Mar 6, 2018
Copy link
Contributor

@whot whot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I found this: https://steemit.com/utopian-io/@tingping/create-script-to-generate-flatpak-manifest-from-pip-packages and that script has been merged into flatpak. So I guess this answers whether it's allowed and the answer is yes :)

Whether it's better to rely on pypi cache or direct git archives I don't know. My gut says rely on git where possible because that's the canonical source of development (AFAIK there's no guarantee that the pypi package matches the git content).

for pyxdg I don't have a good solution other than pypi, but gobject and svgwrite could pull the packages directly from git.

org.freedesktop.tuhi.json Show resolved Hide resolved
"url": "https://github.com/mozman/svgwrite.git"
"type": "archive",
"url": "https://pypi.python.org/packages/a6/e1/8d592fc801e1dc2958fe0c84c733ed729d4020daa1826c58978f9d601bb4/svgwrite-1.1.12.zip#md5=05780a4a8ba33c16842faf37818d670e",
"sha512": "cafce1a39c932149cd9d531c7304e8568fde9bc5abd09b548a778cc7f7f4d74327da03fb01279a3a19dd8396487c8fba974d50c6dd645be3942a4ba6d76646df"
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think using the github package is better here than the pypi cache, removes one layer of indirection:
https://github.com/mozman/svgwrite/archive/v1.1.12.tar.gz

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't we have issues with dates when pulling these tags from github?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we used to but maybe not anymore? I pulled the svgwrite tarball yesterday and today and I'm getting the same md5sum for them.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Homebrew/homebrew-core#18044 (comment) is relevant here: basically, yes, they can change but they shouldn't, at least not that often :)

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

IIRC things attached to a proper release don't change (libwacom e.g. adds the md5sums to those assets too) so we should be able to use those for svgwrite at least - that link above.

@bartsch
Copy link
Contributor

bartsch commented Jun 4, 2019

Do we have a conclusion here? Use archives from github if possible - like we do for pycairo?
pyxdg has an official github-mirror as well: takluyver/pyxdg

@whot
Copy link
Contributor

whot commented Jun 4, 2019

@bartsch eh, given it's been a year, clearly no-one feels strongly enough about it. So feel free to do what you think is the best solution :)

@bartsch
Copy link
Contributor

bartsch commented Jun 6, 2019

@bartsch eh, given it's been a year, clearly no-one feels strongly enough about it. So feel free to do what you think is the best solution :)

Ouch! flatpak-builder requires multiple hundred mb of additional dependencies? I would postpone it until we have something "real" to build.

@whot
Copy link
Contributor

whot commented Jun 6, 2019

anything flatpak pulls in the SDKs etc which come in at a big size. That's by design because that's what makes flatpak work as i. But the theory is that if everything [1] is built with flatpak, you have all that locally anyway. The SDKs don't update much, so once you have them, they'll stick around forever.

[1] fsvo "everything"

@whot
Copy link
Contributor

whot commented Jun 7, 2019

well, turns out at least pygobject has different checksums on pythonhosted vs GNOME's ftp server. Not very trustworthy, the whole thing...

@whot
Copy link
Contributor

whot commented Aug 27, 2019

I'm closing this. This has already cost us more concentration time than pulling the things from git would take over the next year together so let's just go with the git version and move on.

@whot whot closed this Aug 27, 2019
whot added a commit to whot/tuhi that referenced this pull request Aug 27, 2019
flathub requires us to use specific versions, so let's do that here too so
we're consistent in testing.

Fixes tuhiproject#106

Signed-off-by: Peter Hutterer <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Further information is requested
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants