@davefarley77 I'd notice if the style on your videos was different, but wouldn't know it was your video just from the thumbnail style, if that makes sense!
Some text with the title and a few logos/icons/cartoons is >50% of the results for "test driven development" on YT when I search.
@davefarley77 I'm torn! I generally dislike the trend of "face on video", but here it's the difference between "Ah, Dave's published a new video!" and "I have no idea which channel this is from at a glance".
I wonder if a larger CD logo to make up for it would be an idea?
@trisha_gee I went for writing tests, both as a "I spend most of my time doing this", but also as it's the hardest part.
I often work with image data, and designing a test that asserts *something* useful, but isn't just "assert image is identical to an earlier run", is quite challenging.
@trisha_gee My answer would differ quite a bit depending on what cycle you're thinking of.
"Run tests locally" cycle is >75% and seconds.
"Run tests on CI infra" is ~50% and a couple of minutes.
"Run tests, build, and deploy" is <25% and up to 20 mins (which is what I answered for).
@oejwing "HazelFLOOD is designed to answer ... what will happen on site when [a] storm of intensity X occurs?"
So... they're selling flood maps for individual events (that the customer specifies?) at individual sites? How terribly innovative...
@whisperity@GergelyOrosz With the differences that:
1. I'm more confident that MS/Google know how to handle that security
2. Also more confident they wouldn't consider a lack of security a trivial "tech debt" issue
3. They are likely audited to widely-accepted standards, which some customers will require
@pvukovic@GergelyOrosz Presumably, they're currently saying "we don't store your data", and so have absolutely no stance on privacy of stored data or your rights.
But how can you be convinced that they do store nothing, and so be convinced their lack of security for stored data isn't a problem?
@davefarley77@NotMyAgile Looking at the very serious replies from yourself and other big names... we are all aware this is a (brilliant) satire account, right?
@GergelyOrosz From what I've heard from Dutch rail employees (as of yesterday), it sounds as if even internally they understand very little about the underlying cause. About the best they could do is "We are still investigating the problem" style updates.
UK based EO/geo folks - we're hosting a STAC meetup at SatVu's offices - come along! Great views, great company, great metadata discussions #STAC https://t.co/H4q0W9ciDF
@jesusmnavarrol@tottinge Not a big issue though. Very easy to design the majority of tests to use in-memory data by isolating infrastructure from logic, and the library I'm often working with for big number crunching (GDAL) actually has in-built support for its own in-memory filesystem.
@jesusmnavarrol@tottinge I've briefly looked at that in the past, but found:
1. A solution on one OS is unlikely to work on all
2. All solutions seemed to require admin powers, at best requiring a password prompt to run tests, and at worst impossible (for orgs where admin power isn't universal)
@tottinge I guess that really does date me! The idea that you'd have no filesystem tests *at all* was so alien that I would never think of it, so I assumed you meant "just use the filesystem for all tests, it's fine".
@tottinge Ha, I was thinking in Python terms too - life is always slow for me!
Was your original thought hinting that some people would totally avoid filesystem tests, to the detriment of testing the filesystem functionality at all?