#18417: Tarball download fixes
-------------------------------------+-------------------------------------
Reporter: vbraun | Owner:
Type: defect | Status: needs_work
Priority: blocker | Milestone: sage-6.7
Component: build | Resolution:
Keywords: | Merged in:
Authors: Volker Braun | Reviewers: Thierry Monteil, Leif
Report Upstream: N/A | Leonhardy
Branch: | Work issues:
u/vbraun/tarball_download_fixes | Commit:
Dependencies: | 53cc3c352d150b3f2fe9cc342577b809fe60fbc3
| Stopgaps:
-------------------------------------+-------------------------------------
Changes (by leif):
* status: needs_info => needs_work
* reviewer: => Thierry Monteil, Leif Leonhardy
Comment:
Replying to [comment:5 vbraun]:
> I'm just removing the duplicate checksumming, that does in no way
introduce an additional dependency. Old-style spkgs still can't be
checksummed because they don't have a checksum.
I was wondering whether you didn't know what you're doing or whether you
just didn't care... (and I didn't want to further comment on the second
commit ''here'' in hope you'd [re]move it)
The '''second commit''', which is completely unrelated and independent of
the first one (I would have given positive review if you [re]moved the
former from this ticket, where it simply doesn't belong), needlessly
introduces new requirements for building Sage, even from the (self-
contained) source tarball:
* a system-wide Python installation (with a working urllib)
* internet access (to github btw.)
* write access to the `$SAGE_DISTFILES` folder (cf.
[ticket:15642#comment:85 my previous comment]) -- ok, the build
'''currently''' doesn't fail otherwise (see below), but `sage-download-
file` keeps downloading the mirror list for each and every package subject
to installation, continually re-ranking the mirrors
and all of that '''regardless''' of whether any package actually needs to
get downloaded. (Thierry already mentioned silently incorporating such
changes into a ''blocker ticket'', and late in the release cycle, is even
more dubious.)
[[BR]]
That the build without a system-wide Python and/or without internet access
'''currently''' still succeeds is just '''due to a bug''' in `sage-spkg`,
namely that it doesn't at all check the exit code of `sage-download-file`.
So as is (with your second commit included), checksum errors wouldn't
necessarily lead to (build) errors, hardly anybody would notice them.
''In case Python is available'', and your script detects the checksum of
the tarball already present in `$SAGE_DISTFILES` doesn't match the one in
`build/pkgs/...`, ''it deletes the tarball'' (provided write access is
granted) ''without asking'' and tries to download it. (With
`$SAGE_DISTFILES` read-only, `sage-spkg` wouldn't notice checksum errors
and simply proceed with the "wrong" tarball; same in case Python wasn't
available -- silently no checksums would get verified, until Sage's Python
has been built, but then we're back where `sage-dist-files` only ''tries''
to delete "wrong" tarballs...)
--
Ticket URL: <http://trac.sagemath.org/ticket/18417#comment:8>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/d/optout.
