The propsect of Paramount settling to Trump for 60 Minutes editing a Kamala Harris interview is alarming (1). This seems to ultimately boil down to the inherent conflict of business pragmatism and journalistic integrity. This is largely implied within the article, but is also a clear example of that concern and how it may have wider ripples as corporations get in the ring-kissing line. I'm particularly concerned about it given that I've really come to appreciate the value of a free press over recent years, and one of my potential "deal breakers" would be signs that relatively independent information is no longer available (since then you've ceded the ability to make an informed decision).
While I certainly don't have a deep legal background, not only does the described suit seem to have questionable primary merit, it would also presumbly hinge on establishing proof of harm. While I haven't seen the clip in question it certainly is not described as having content that would have direct negative impact on Trump (and without such proof he should lack the standing to bring the suit), and therefore this seems to amount to allowing our legal system to be dragged down in to playground petulance.
I certainly understand the multiple pragmatic reasons to settle from a business perspective, but it also seems to pose a compromise in a field which should cling to principles and a dangerous precedent.
A few weeks ago I was working on getting some python libraries loaded into an image using the Bazel Python rules (2). Likely the biggest challenge is that story of packaging in the Python ecosystem seems to lag behind many others (there's a blog post by Alyssa Coughlan that I read many years ago that outlines some of the background).
A particular challenge was that I wanted to pull in existing artifacts without building anything new - which isn't so much a distinct challenge but is not the focus of the documentation.
Creating a requirements.in file which could be compiled
to a requirements.txt file all clicked together once I
dusted off some dormant Python knowledge. The actual loading of the
files into the image in a desirable form took a little bit of fiddling.
The end goal was to retrieve the desired packages and any transitive
dependencies. Using py_library and selecting the desired
requirement allowed me to pull in all the relevant files in an expanded
form, ready to be consumed by the Python runtime.
For use of the package I opted for a small shell script which would
set the PYTHONPATH to the place where the above files were
placed before invoking python -m with the desired package
(where the locations are all coordinated by the bazel file using things
like environment variables. This offers the advantage that different
applications could make use of different sets of Python packages
(analogous to venvs).
One additional wrinkle was around any native extensions - the bazel
file was written to produce multi-platform images and, among other
challenges, pip seems to be focused on building for the
current host without clear support for cross-compilation. The workaround
for this was to install system (Alpine) packages for impacted libraries
and prepend the host location to PYTHONPATH. This
definitely feels a bit rough around the edges but gets the job done.
In the longer term I'd want to dig more deeply into the Python side of things to refine some better packaging practices where bazel just provides some glue…but at present I'm not actively working with Python so this will just be left as is.