morgenkaffee

Yes, this dataset is not CC BY-NC 4.0. Be careful.

Well done. Every conpany with maps otherwise needs their own daily diff server.

The challenge tho is running it over years reliably especially if something in upstream OSM breaks.

I use Docusign cause it looks official and grown up. With this product I worry about the name and domain name looking like phishing or a typo.

I am a fan of replacing docusign as a principle though.

I also found it hard to deal with the AWS CLI and Parquet. I built this site so you can download it in CSV and Geopackage.

https://repromptai.com/data/foursquare

Yeah there's lots of tuning to do on the commit messages (like filtering out bots). I really believe by having access to git diffs and commit messages you can fully automate a standup.

These billing changes make some self hosted tile users go from $0 to tens of thousands of dollars a month in usage. It’s shady, don’t lie to yourself.

This only happens if the developer deliberately upgrades and adds a token. If you do nothing and you keep using v1 - nothing changes for you.

I really want to keep it a standalone style editor. The advantage is exactly that it is not bound to any external service.

No not really. The goal is to stay simple - just display polygons, linestrings, points and their associated metadata. It is really meant for prototyping SQL queries quickly with immediate feedback.

What might make sense is to support to display several SQL queries at once. So you can show the the objects within a cluster and the cluster bounds at the same time.

For complex visualizations you can use e.g. QGIS with PostGIS as data source. I used PostGIS editor before to prototype clustering SQL queries. I assign cluster values as attributes and inspect them like this.

It is all magic of Clang or GCC. Once you call the template with T1 the C++ compiler will enforce that your passed type supports the <= operator.

Yes you would have to program in a subset of python. But type annotations are not always needed.

For the array I do a little hackery. You can define the array without an initial value in the container and I can guess the value type.

  arr = []
  arr.append(1)

  std::vector<decltype(1)> arr{};
  arr.push_back(1);

You are right. C++ is quite different to Python and so it should be called a compiler.

But it fits the project since I don't have any background or experience in compiler design.

I can only encourage to use Haskell for such kinds of problems. This was my first Haskell program and I was productive quite fast thanks to "Learn you a Haskell" and the great Parsec library for parsing the Dockerfile.

If you are an experienced Haskeller I would be very glad if you look over the project and problems in an issue. Because mostly I have no idea whether what I am doing is idiomatic or not.

Completely agree!

One approach is to run it in a Docker container and access it from your other micro services and never expose it to the public. This is still handy because you can isolate that old program and act as if it is an API to your other micro services.

I don't have any experience with zeromq as I mostly deal with AMQP protocols. But good that you mention it - if there is a demand I can dig into it.

There are quite a few other message queue systems out there. I can also imagine that pipecat could work as a kind of migration systems between them. But this probably should be a separate tool as it requires shuffling around metadata as well.

Now that I look at it again it also seems offensive. Switched it to a more boring but more suitable logo.

Thanks.

Sorry for the inconvenience. I moved the install from source section up in the README.

Perhaps there is a better way than providing a download from github releases and chmod+x it?

I know you can already do that with the `COPY FROM` statement in Postgres but I always think it is a pain because you need physical access, need the files to be 100% correct and you have no progress indication.

If you need a more sophisticated solution you can look at pgloader. http://pgloader.io/

Ah that's the same idea! Wished I found that before. Thanks man.

Yes you're right. I can create a flag --publish and --subscribe.

Mine is even more poor :)

I use it as a poor man's job queue.

cat jobs.txt | redis-pipe jobs

An then on several workers: redis-pipe --count 10 jobs | python do-work.py | redis-pipe results

An then in the end: redis-pipe results > results.txt

Or you can use it as simple logging mechanism: tail -f /var/log/syslog | ./redis-pipe logs