@otfrom It doesn't solve all the troll problems, but it limits them to something manageable.
Anonymity can be reached even with small instances or personnal instances on a federated network. After all, the fediverse as a whole is large enough so that a single user is relatively anonymous.

@otfrom I did that kind of remark in the past about the fediverse : each instance should be limited to a small community of users, around 200 people max. For moderation, I read that there is usually only a very small percentage of toxic users in communities, I remember around 5%. For a community of 200 people, that's around 10 people, so that's manageable by one or two moderators. For a community of 2000 people, that represents 100 people, and it's harder to moderate by small teams of moderators. Mastodon.social has 706K users (105K active users), so potentially, it has 35K toxic users (5K active toxic users). So it's impossible to moderate.

@juliobiason Those so-called header-only libraries are just full of inlined methods and functions.

re: Some thoughts on programming 

@loke The horrors of programming I have encountered and endured are more often linked to bad management practices and toxic behaviours than to the development work itself. The presence of programming anti-patterns is often a good indicator of management issues (understaffed teams, high turn-over, too short or impossible deadlines, lack of maintainance, lack of directions, lack of long-term vision, everchanging priorities, lack of trust....).

@schaueho The release model of Debian is to blame here.
The reputation of being obsolete when released is really dangerous when it comes to security, as many open-source software don't have the man-power for the long-term support that Debian wants.

@juliobiason Probably because there are different (and incompatible) use cases behind the mpmc idea.
When a message is produced, should it be dispatched to all consumers (publication-subscription on a topic), or just to the first idle consumer (load-balancing) ?

@MutoShack Sometimes, the environment in which you must test the software cannot be accessed from home. Notably when the software is embedded in a system.

@mkf but no tag 0.13.0 on that repository, so linux distributions won't take it.

@mkf For pilot-link, though I can't find the version 0.13.0. The website of pilot-link is down (name resolution failed). I've seen there is a deb package distributed on the jpilot website, but without a source archive, I can't really flag the package out-of-date.

@mkf For ArchLinux, the packages being in AUR, it's up to the users to flag them out-of-date. Also, AUR is not maintained by the distribution, but by other users, so they may just not have noticed a newer version came out.
I'll flag the packages on AUR since I can do that, but I can't assure they will be updated quickly.

@tfb extending the flex office capacity of the building by using the parking

@juliobiason That's the problem with C++, its standard library is rather small, and there are no standardized serialisation libraries.
If you don't know which library to use for parsing JSON, I can suggest jsoncpp, it's rather easy to use.

@juliobiason Boost in a nutshell.

(Well, not that true, some of the libraries also have source files).

@MutoShack Which one should I use ? Fountain pen ? Ball pen ? Felt pen ?

@loke If you have a way to select function bodies based on the value of arguments, you don't need a conditionnal construction. I don't know if this applies to APL though.
Erlang does not have conditionnal, but you can define functions where arguments are partially evalutated, so you can define factorial(0), factorial(1) and factorial(n) with separate bodies. You can simulate a if with a fonction where a boolean argument selects the body to use (you can also use guards to select a function body).

Show older
Functional Café

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!