Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
tyingq
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
47 ms
·
841.
▲
by
tyingq
3y ago
Google's pretty fond of privacy things that "hurt advertisers", but hurt all advertisers other than Google harder than they hurt Google.
842.
▲
by
tyingq
3y ago
Better...it's also an "early warning sign". Oof.
843.
▲
by
tyingq
3y ago
I'm with you on the idea that mainframe COBOL is typically only part of the actual program. But COBOL not being Turing complete sounds...not right.
844.
▲
by
tyingq
3y ago
The ones I've worked with weren't solely just if->then->else trees. There were also things like weightings, probabilities, risks, etc.
845.
▲
by
tyingq
3y ago
Philip K. Dick guesses right again. Pre-crime.
846.
▲
by
tyingq
3y ago
I suppose I'm wrong because it appears not to exist. But, I can't get past the idea that there's just a lot of things very few people have more than one of. Many of them being big ticket items too, where the cost (or opportu
847.
▲
by
tyingq
3y ago
That's a good point. Now I'm curious if they have some internal tagging that differentiates between these two kinds of products.
848.
▲
by
tyingq
3y ago
I see language in Amazon's ad seller pages like: "We will always automatically exclude recent past purchasers of your promoted products." So there's some notion of this data being available, at least internally.
849.
▲
by
tyingq
3y ago
For an idea of that overhead, see this older article: https://hacks.mozilla.org/2018/10/calls-between-javascript-a... At first, it sounds really promising, like "wow, all the overhead is fixed". Until
850.
▲
by
tyingq
3y ago
I mean it boils down to oil and presence/influence in areas close to the oil.
851.
▲
by
tyingq
3y ago
Amazon still re-shows toaster ads when they know for certain you've bought one. Maybe the likelihood you're buying one for someone else as a gift is higher than their guess of what else you might buy?
852.
▲
by
tyingq
3y ago
> Is there something in the awk script that makes it advantageous over a shell script? Pseudo multi-dimensional associative arrays for representing the dependency graph of make. This part: for (i = 2; i <= NF; i++) slist[nm
853.
▲
by
tyingq
3y ago
Canadian maybe? https://en.wikipedia.org/wiki/Alex_(videotex_service)
854.
▲
by
tyingq
3y ago
Making them one product seems possible/reasonable. Cloudflare does it.
855.
▲
by
tyingq
3y ago
The syntax takes a little trial and error and usually finding real-world examples, but I like "make". I had one project that involved downloading a jdk, using it to build a project specific jre, patching some class files with loca
856.
▲
by
tyingq
3y ago
Reads like it's not real straightforward or well documented for rust: https://github.com/nginx/unit/issues/738
857.
▲
by
tyingq
3y ago
Has me curious if it was a scope thing where there was, at one time, a locally scoped print()...or just the usually-writes-python guy was trying to debug something and forgot to remove that.
858.
▲
by
tyingq
3y ago
A fair while ago, I had luck with motion activated sprinklers for raccoons. I would think they would be effective for cats also.
859.
▲
by
tyingq
3y ago
Sounds like the idea is that the fake profiles are created so they can make mass "friend requests". Then, when those friend requests are accepted by some amount of your actual friends, it unlocks more access to your data. Or ev
860.
▲
by
tyingq
3y ago
Doesn't the -A zip cause it to skip the subpages? That's what I'm saying...the filter is keeping you from traversing main page -> sub page -> zip file.
861.
▲
by
tyingq
3y ago
If they believe their own assertations, like: - that it doesn't output training data verbatim - the product is very transformative, only "learning" from training data - There are no copyright infringements because of these tw
862.
▲
by
tyingq
3y ago
They have the ability to feed their closed source to Copilot for training without exposing the source to everyone directly, given the relationship. They choose not to.
863.
▲
by
tyingq
3y ago
Ah, so for copyright reasons.
864.
▲
by
tyingq
3y ago
Yet they don't feed their own closed source assets to Copilot for training...why not?
865.
▲
by
tyingq
3y ago
Pretty sure the payments were to individuals, and not to any Rotten Tomatoes entity directly. So you would have to make some case about lack of due diligence instead, or sue a bunch of random people.
866.
▲
by
tyingq
3y ago
Makes me curious what their revenue trends look like. From the outside, it feels like many people in the space are using the Arduino IDE, but with MCUs they bought from some other vendor.
867.
▲
by
tyingq
3y ago
Ah, yes I read that. The wildcard would be "undisclosed terms"...if there was some restriction of not directly owning/running a record label.
868.
▲
by
tyingq
3y ago
Curious if Apple buying a record label somehow re-stirs the Beatles/Apple Corp/Apple Records vs Apple Inc fight.
869.
▲
by
tyingq
3y ago
I think that would work if that page directly had references to the zip files, but it doesn't. The references are on the sub pages for each font.
870.
▲
by
tyingq
3y ago
The mirror option depends on a hierarchy where it can exclude parents. Since the actual downloads are on a different domain, it seems this won't work. A hacky script to do it: #!/bin/bash for page in $(curl -s https:&
More ›