-
-
Notifications
You must be signed in to change notification settings - Fork 7.9k
Download Qhull source from Github, not Qhull servers, in meson build #27505
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
Download Qhull source from Github, not Qhull servers, in meson build #27505
Conversation
Co-authored-by: Elliott Sales de Andrade <quantum.analyst@gmail.com>
What is included / excluded in the download tarball vs what we are now getting from github? |
I'm not familiar with exactly what command Qhull used to create their md5 from the tarball, but with
+ some python to sort the list, I was able to git diff the files. md5 sums from github tarball
This doesn't exactly match the md5 file listed on qhull's download page. Most of the |
Oh, I see I took the qhull tarball "qhull-2020-src-8.0.2-tgz" to mean git tag 8.0.2, but I also see there's a git tag for 2020.2. I'll check that one. EDIT: Looks like both the github tarballs from the 2020.2 and 8.0.2 release match each other exactly, and neither matches the one on qhull's webpage. Of course, I'm using their stated md5s, since I can't download the tarball from their website, if you can, the simplest one to check is |
Beyond the superficial filepath (one had a subfolder of Most of the MD5 differences come from kyle@radon py@edge Downloads $ diff -r --strip-trailing-cr qhull_org github
Only in github/qhull-8.0.2/build: qhulltest
Only in github/qhull-8.0.2/build: README-build.txt
Only in github/qhull-8.0.2: .gitignore
diff -r --strip-trailing-cr qhull_org/qhull-8.0.2/html/qh-get.htm github/qhull-8.0.2/html/qh-get.htm
96c96
< <li>Sept. 3, 2020 22:00:41 EDT md5sum for Qhull 2020.2 downloads
---
> <li>Sept. 3, 2020 22:14 EDT md5sum for Qhull 2020.2 downloads
98,101c98,101
< 368f2aa9bf89677af04d57d281041359 *qhull-2020.2/qhull-2020.2-zip.md5sum
< 27c44a7e2db5f127e1c6fc7807580566 *qhull-2020.2/qhull-2020-src-8.0.2-tgz.md5sum
< a1a58969aa17d89088ebbd0b0eb44a82 *qhull-2020.2.zip
< fa34047579937433e3a36c5d4033a988 *qhull-2020-src-8.0.2.tgz
---
> 4f68ebb4700849a548b332de500d42e9 *qhull-2020.2/qhull-2020.2-zip.md5sum
> ed767244864488d5e9a22dfa788663a9 *qhull-2020.2/qhull-2020-src-8.0.2-tgz.md5sum
> 7ed0bcce474bf12f6fbb5b4093760f3d *qhull-2020.2.zip
> 295f7332269a38279478f555cc185296 *qhull-2020-src-8.0.2.tgz
Only in qhull_org/qhull-8.0.2: qhull-2020-src-8.0.2-tgz.md5sum So the only file which differs in actual content is the one listing the md5 sums of the tarballs/zips/md5sum files. The qhull.org version is the missing The github version is missing one md5sum file. None of these seem problematic to me |
Hi, I am currently failing to build matplotlib 3.8 using pip install, because again it seems the old qhull website has some issues and the connection is reset in the middle of the download. I see this was merged for meson only, which seems to be only in 3.9. How possible would it be possible to backport an equivalent of this, but in 3.8 with the previous build system ? |
Change similar too matplotlib#27505 but on the branch 3.8.x as right now the qhull website seems unable to properly deliver the tarball without disconnecting in the middle.
Change similar too matplotlib#27505 but on the branch 3.8.x as right now the qhull website seems unable to properly deliver the tarball without disconnecting in the middle.
PR summary
Closes #27159. As mentioned there, the Qhull server is HTTP (not HTTPS). While that's not a security problem in theory (we know the hash to expect), in practice corporate/academic may servers block the download. This includes eduroam, used by many large research universities. Furthermore, as @tacaswell mentioned, rather than hitting Qhull's server with every build, it's better to rely on github. (IIUC - lmk if I'm misrepresenting!).
The original issue got a lot of attention, but appeared resolved when I moved to a new network. When I met a new colleague who mentioned the same issue, I reopened the issue - understand if this needs more discussion beforehand.
PR checklist