• 6 Posts
  • 942 Comments
Joined 2 years ago
cake
Cake day: June 12th, 2023

help-circle


  • The few things I don’t like about flatpaks (which become a problem on atomic distros that use almost all flatpak by design):

    • Some types of embedded development is essentially impossible with flatpaks. Try getting the J-link software connected with nrftools and then everything linked to VScodium/codeoss

    • Digital signing simply doesn’t work, won’t work for the foreseeable future, and is not planned to get working,

    • Flatpaks sometimes have bugs for no reasons when their package-manager counterparts don’t (e.g. in KiCAD 8.0, the upper 20% or so of dialog boxes were unclickable with the mouse, but I could select and modify them with the keyboard, only the flatpak version)

    • The status on whether it is still being actively developed or not (at least I hear a fair amount of drama surrounding it)

    But besides those small things, it seem great to me.


  • Some drives are worse than others and higher capacities get worse and worse, in my experience, Seagate drives are extremely loud.

    If you get helium drives (like wd red plus > 8TB i think),or 2nd hand hgst/ WD enterprise drives) they are significantly quieter.

    But, having an ssd is cheaper probably. I have an SSD for the boot drive and all databases, configuration folders, etc… In docker so general IO is fast, then media, documents, pictures, etc… On the big HDDs.


  • Absolutely can’t wait for new battery tech for grid storage too! Sand batteries that can use otherwise-unusable sand, sodium-Ion batteries (or mainly inverters that can handle the expanded voltage range compared to Lithium-based), expansion of pumped water batteries where it works. This is about to be THE time for government-funded alternative batteries across the world. Energy would get so plentiful that it wouldn’t even be profitable for fossil fuels anymore. That is the dream. Of course there is a 99% chance that every single government in the world drops the ball completely.



  • KNX.

    Everything is decentrally programmed, and you can do extra automations and stuff from home assistant, but KNX devices are wired (generally) and will always Just Work™. More expensive that the cheaper retrofit options, but if you factor in manual overrides or getting the “better” wireless smart devices it is comparable. They generally also have a manual override at the panel. For core functions like lights, HVAC, roll shutters or blinds, etc… That is honestly the best option (unless you want every light to be an RGB light for some reason, then you still need smart bulbs)






  • London population: 8.8 million

    Twin cities combined population: 3.6 million

    London public transport: pretty damn good, connections everywhere, not an insane price

    Twin cities public transport: almost non-existant, insane parking prices

    London police: sometimes reasonable and lightly armed

    Twin cities police: notoriously corrupt, heavily armed, use constant excessive force on civilians

    Mystery solved.





  • True, but I have never seen an ECAD software that has implemented that because that opens up a whole other can of buggy worms of intentionally trying to connect nets and guessing the intention of the user. When do you push the net and when do you merge the net? Does dragging a net label connected to a net to intentionally attach it to a second net to connect the two not work due to pushing? When do you decide when dragging a net is an intentional connection vs not during the drag? When you drag a component, do all nets and connected components get dragged with it or do they push each other into a jumbled mess of traces jumping over each other like what sometimes happens with push behavior in layout routing? I think it is a pretty difficult problem to solve.

    That’s why I personally like KiCAD’s choice just to not pull nets with component moving, but to each their own.


  • Nope, I mean in schematics. You can absolutely short two lines together. It will just silently merge them into a single net so it won’t show up in ERC. If you have an IC with 8 pins on each schematic symbol side in altium and drag the component one unit down, 7 and 7 pins will be shorted together (as an example) silently and become two nets.

    As far as footprints:

    You don’t technically need to, but altium’s spaghetti code makes things break so often and my company had some troubles with that so they recommend making a duplicate every time.

    Not to mention that searching for the correct footprint is accomplished 3 different ways and each way is partially or fully broken so you pretty much need the exact sequential footprint symbol number to sort and scroll all the way to find it. It is horrific. No organization at all.



  • That is true.

    However, the alternative is altium behavior where it drags all of the wire connections with you, so if you move anything attached to an IC or the IC itself, you get dozens of shorts immediately.

    They both have pros and cons. I actually prefer kicad’s way because it will never lead to unintended un-ERC-discoverable shorts.

    I have used both KiCAD and Altium regularly for years and there are many things that KiCAD simply does better but it is missing a ton of QoL things.

    The one thing that I don’t like about KiCAD is that some shortcuts don’t have an alternative right click or toolbar menu item, which makes them undiscoverable unless you browse shortcuts.

    I really like librePCBs approach to library management. Multiple pinouts for schematic symbols (meaning a BGA and QFN can have the same library item) and the categorization.

    Though I can’t tell if they have reusable footprints and are able to simply reference them to a schematic symbol which is one of the nice things in KiCAD over altium