Hmm, that's a bit concerning if too long posts don't federate because of a server-side limit. I mean, ok don't federate a 100TB post and destroy my instance by eating up all the disk space, but if an instance admin wants to support the 12k character limit that Mastodon has, that should be an option that the admin has access to configure.
abff08f4813c
joined 11 months ago
Ah, I just saw https://codeberg.org/rimu/pyfedi/commit/e71f4dbb92f63f3da285707cfb418d115a4de226#diff-e919f1e867cfa561970ba9549250f4d5bb90bcf6
That's pretty cool - amazed also at how quickly you're able to get to these improvements. Keep up the good work!
while Beijing is still committed to taking control of Taiwan, it’s hesitant to start a direct war.
Only "hesitant" ? It's a bit reassuring to hear that a war isn't wanted ... but it would be better if war was renounced in stronger terms, especially because
they create a serious risk of miscalculation that could spiral into a larger conflict.
Agreed, I can easily see that there are things here and there that no one has had time to address or perhaps even an oversight or two regarding a potential bug. I believe that your feedback is valuable and helpful. We should always be on the lookout on how to do things better, and open to discussion on it.
Same here.
It’s kind of an issue though if the documents cannot be understood by anyone but the lead dev.
The one you mentioned is an exception though, most of the docs are quite easy to understand. Examples:
https://codeberg.org/rimu/pyfedi/src/branch/main/docs/ADDING/_COMMUNITIES.md
https://codeberg.org/rimu/pyfedi/src/branch/main/docs/ADDING/_USERS.md
https://codeberg.org/rimu/pyfedi/commits/branch/main/docs/KEY/_TERMS.md
https://codeberg.org/rimu/pyfedi/src/branch/main/INSTALL.md
https://codeberg.org/rimu/pyfedi/src/branch/main/INSTALL-docker.md
I suppose a weakness is that when it comes to the code itself, most of the explanation is on youtube rather than being written down, see https://join.piefed.social/2024/01/22/an-introduction-to-the-piefed-codebase/
It looks to me a lot like those non-profit founders who explain nothing, make contributing hard,
This reminds me of kbin. But I don't think it's quite fair to paint rimu the same way - he goes out of his way to help folks get set up and to encourage development.
They say “help wanted” but if you believe it and contribute you’re going to get burnt out.
Again, reminding me of kbin. But if you look at the pull requests though, there are many of the same folks contributing over a span of months, mostly with merged PRs. I can't speak for any one of those folks, and burn out generally is a very real concern, but, not quite seeing evidence yet for this project specifically (though again perhaps I've overlooked something).
Seeing the size and amount of features though, it’s already past prototype stage imo.
When you don’t use one, you end up with a different code organization imo, and different affordances (eg more SQL views, several specialized data classes, a db layer that returns exactly what the program needs…)
You always end up creating half a framework anyway ... and so the questions are 1/ is it needed 2/ do you feel like designing one?
These are good points. They remind me of the old debates about Minix vs Linux, the kernel. My guess is that since pyfedi is meant to be easy for beginners to understand and modify on their own, that this influences some of the design decisions made. But of course, even here, reasonable folks can reasonably disagree on what's best.
Honestly what irked me was mostly a succession of operations that weren’t in a transaction , and I also wondered why that wasn’t a trigger or a hook.
Yeah, I don't really know about this one. Your suggestions here make sense and wouldn't clash with the "keep it easy for beginners" philosophy, so off the top of my head I can't think of a reason why it was done differently - though perhaps I'd be able to understand it better if given more context.
Agreed. In addition, I'd add this to the OC's comments:
It would be like if you were at a pro Palestinian protest and were asked to sign a document saying you’re free to protest but only if you remain quiet about certain things the isralie government is doing
Well, no. It's more like, the Israeli gov't promises that you can get secret information from them about what they are doing, but you can't mention it publicly until they've made it public. But worst case you're still free to protest anything and everything that's already public.
I'm not sure if even that premise is correct, though. From https://www.theglobeandmail.com/politics/article-poilievre-is-the-sole-party-leader-foregoing-access-to-classified/ (archive https://archive.is/QEbVP)
top intelligence officials have said that secrecy rules would not prevent leaders from acting on the information they receive. On Tuesday, Green Party Leader Elizabeth May held a lengthy news conference detailing what she learned from the classified report.
Likewise, the previous opposition leader did have the clearance, and was able to speak publicly about what he learned, as per https://globalnews.ca/news/9732593/erin-otoole-chinese-interference/
Former Conservative Leader Erin O’Toole says the Canadian Security Intelligence Service (CSIS) found an “active campaign of voter suppression” by China against him and his party in the 2021 election. O’Toole made the comments Tuesday from the floor of the House of Commons
So this excuse rings hollow.
not because of anything you did but because you refused to sign the document.
A refusal to perform a certain action is a choice. And choices have consequences...
If you refuse to sign, a crowd of people all start shouting that
I haven't seen anyone call PP a Nazi specifically because of this, yet. So I think the analogy breaks down here.
Also, signing the document is just a step, it's not all there is. Even though PP had a security clearance in the past that has since expired, presumably he'd have to be rechecked before getting a new one. Unlike just signing a doc, there are some concerns here that PP might actually have something in his background that would cause him to fail this check, resulting in a denial even after he agrees to sign.
As an MP He’s a representative not a member of the federal government.
No, the government is the party that governs
Parliament is the legislative branch of government. Its main purpose is to make laws and hold the government to account.
Government is a broader term with different meanings. Inside the House of Commons, it usually refers to the Prime Minister, Cabinet and other members of the governing party.
So that is accurate.
Let the opposition be the opposition.
How come only PP didn't get a clearance? Every other party leader has one. Think of all the times Bloc Quebecois clashed with the Liberals (e.g. https://globalnews.ca/news/10791235/bloc-quebecois-pension-payments-possible-election/ and https://globalnews.ca/news/10791235/bloc-quebecois-pension-payments-possible-election/ ) but their leader still has the clearance. Doesn't seem like having a clearance hurt that much.
I can't find confirmation but I imagine the last time that the Liberals were the Opposition, their leader, Michael Ignatieff, would have held this clearance as well. So PP is very much the odd one out here.
Good related discussion on this over at https://lemmy.ca/post/41338469
I suspect that this is just the tip of the iceberg and we're a) going to be seeing a lot more of this publicly in days to come and b) there's going to be even more where folks do the move but try to keep quiet about it.
Glad to hear your daughter-in-law and her mom are safe. Fingers crossed that the brother is as well.
Edit: hey I found this in the architecture document:
What I tarnation is that project? This lack of test is literary on purpose?! Also reliability and security are this low?
Only rimu can answer this. Note though that the architecture document you are citing from has only a single commit and the doc itself states that it's a work-in-progress. So these are likely not binding guidelines, if someone wanted to add more tests and had a PR ready then I'm sure the devs would be open to discussing things.
I'm also half-wondering if the scores don't actually reflect rimu's view of their importance to the project - but merely his view on how well the current codebase implements those features. Or perhaps just his own time commitment.
why use a micro web framework when clearly the scope is a better fit for Django, and the devs don’t have the knowledge or time or energy to constantly refactor to keep the complexity under control?
If it were up to me I'd probably keep it as Flask. As per https://www.greengeeks.com/blog/django-vs-flask-python-framework/
Flask is known for its simplicity and flexibility
Some argue that Flask enables faster development due to its minimal setup and flexibility
This fits with my personal experience with both frameworks. Also keep in mind that rimu (as he states in the screencast in the README) wants this to be a project that is easy to build on top off and contribute to for folks who are newer to software development - my understanding is that the goal is to lower the technical bar so folks can get hands-on experience with developing for the fediverse more quickly and more easily.
no tests?!
Along those lines, adding unit tests can be a good way for someone new to a codebase to become more familiar with it.
many 1000+ lines long files of python, a high level language
But if you look at it, it's mostly that there are a lot of small methods grouped together into a file, see https://codeberg.org/rimu/pyfedi/src/branch/main/app/activitypub/routes.py#L568 as an example.
Even when dealing with a method that's a big longer - e.g. https://codeberg.org/rimu/pyfedi/src/branch/main/app/activitypub/routes.py#L279 which is just under a hundred lines - they're quite readable and it's easy to see what each part of the method is doing.
Of course they could be split up, and some developers have advocated for that sort of style generally - for example https://thoughtbot.com/blog/sandi-metz-rules-for-developers
I think the current organization on pyfedi is reasonable, but it's clearly not following the Sandi Metz rules.
underused ORM (and I don’t like ORMs much)
Not clear on what you mean by this. pyfedi uses SQLAcademy which is one of the more popular ones out there for python, so I wouldn't call it underused.
Or do you mean that pyfedi underuses SQLAcademy? But then, if you don't like ORMs, wouldn't this be a good thing?
The reputation points shown there are calculated by Mbin
Yup, accurate.
PieFed instances ... [do] not show the karma score publicly in the UI to anyone except admins.
Accurate.
PieFed instances do not share the karma score with other instances
Correct in the sense that the total isn't being shared with anyone.
But upvotes and downvotes* are federated, which is something that I'm still seeing being a surprise to new users, despite it being around since when kbin was young as per https://sh.itjust.works/post/361524 (In fact just had one recently, https://lemmy.world/post/27042619/15815241 )
So we'd expect that (assuming there are no issues with federation) the mbin reputation score would include any upvotes or downvotes made by piefed.social users, for example. So on this thread itself, we can see who upvoted it, https://fedia.io/m/piefed/_meta@piefed.social/t/1970895/comments-on-the-points-in-https-join-piefed-social-2024-06-22-piefed-features-for-growing-healthy-communities/favourites and that one of the upvotes is from a piefed user
- Unless one is on a piefed instance that has private voting enabled, https://lemmy.ca/post/27303908
Ah, my mistake, you are right. I did not look closely enough. fedia.io's reduce tab .. isn't a tab, it's just text. Clicking on it doesn't do anything. Also, it seems to lack the route to show the downvotes (e.g. https://fedia.io/m/piefed/_meta@piefed.social/t/1970895/comments-on-the-points-in-https-join-piefed-social-2024-06-22-piefed-features-for-growing-healthy-communities/down takes one to a 404).
I had to search out actual OG kbin servers (e.g. https://bin.pol.social/ and https://kbin.spritesserver.nl/ ) to double check. So OG kbin does show the downvotes (see for example https://bin.pol.social/m/kbinMeta@kbin.social/t/11031/Hey-that-s-cool/votes/down and https://bin.pol.social/m/wiadomosci@szmer.info/t/19503/Kosiniak-Kamysz-Przez-skladke-zdrowotna-firmy-nie-moga-tworzyc-miejsc-pracy/votes/down ) - oddly though, the only examples I can find list only downvotes from users local to the instance.
Anyways, it seems then that while upvotes are public, no major instance currently shows downvotes publicly, and no major forumiverse software (forum-like software in the fediverse) supports showing it either.
Now ... if you're expecting all votes to be leaning towards privacy, then you might still end up having a problem, in particular if you're upvoting something controversial (for example: https://fedia.io/m/politics@lemmy.world/t/1335132/Trump-overtakes-Harris-in-our-forecast/favourites ) but it's still definitely an improvement for privacy-expecting new users.