kevincox

joined 4 years ago
MODERATOR OF
[–] kevincox@lemmy.ml 1 points 1 year ago

I am not sure theres a legal way to retroactively change licenses and terms?

I don't think this is really true. If you own the code you can licence it as many times and in as many ways as you want. You just say that you are licensing existing versions and it becomes true. Probably a good idea to write that down somewhere (maybe in the LICENSE file) to make a "paper trail" in case you ever need to document this license.

You generally can't retroactively change a license, such as revoking it. But licences may have revocation terms. But it doesn't really matter what existing licences are on a code, you can always add new licences if you own it.

A silly example. If I own some code I can say "Anyone standing on their left foot can use this code". That doesn't stop me from later saying "Anyone standing on their right foot can use this code". But licenses are still valid, and I haven't change the first one. But now people can choose which license they would like to use.

[–] kevincox@lemmy.ml 1 points 1 year ago (1 children)
  1. Can I just add a license after the fact and it will be valid for all prior work?

This is fun question because it hinges on a silly technicality of software development. If you add a license to your repo today, the license applies to the code as of that point in the commit history.

I think you are getting a bit too caught up at the process of adding a LICENCE file to a repo. This is just one way to licence some work. You can just say "I as the copyright owner of this work licence it, including all previous revisions under licence X". That is also licensing the code and doesn't matter if the LICENCE file exists at any particular commit.

But yes, I would say that by default adding a LICENCE file would be interpreted just as releasing that version of the code under the particular licence.

[–] kevincox@lemmy.ml 1 points 1 year ago

In general, If you did work for your employer they own the code. You may wish to just ask them, maybe they are fine with adding a license.

If you didn't write it for them, during "work hours" or using their resources then you may own the code. Some employment contracts will claim that they own everything you wrote while employed for them but that may not be legal in some jurisdictions.

[–] kevincox@lemmy.ml 15 points 1 year ago

Yeah, I did the same. I then proceeded to block them and never purchased anything from them again.

[–] kevincox@lemmy.ml 4 points 1 year ago

I've never noticed an issue here. Looking into it the thread messages default to 1rem, your browser's default font size. So if this isn't suitable for you you may want to change your browser's font size.

The font size in the feed is 1.25x your default font size, which seems fairly reasonable for titles. But if you want to change that you can apply some user styles. For example to make it only 1.1x as large:

h5, .h5 {
  font-size: 1.1rem !important;
}
[–] kevincox@lemmy.ml 6 points 1 year ago

Absolutely. They are sort of a compression scheme so the tokens contain different numbers of characters based on how frequent that string is. So common words like "the" will typically be one token, or maybe even common phrases like "I am". On the other hand rare punctuation such as "~" may be its own token. There will also be tokens for many common prefixes and suffixes such as "non" and "n't". The tokens of each model are different but they definitely vary in length.

[–] kevincox@lemmy.ml 3 points 1 year ago (2 children)

I would actually love this. I use email for everything, it is so nice to have everything come to the same place. Right now I follow a few Mastodon users via an RSS-to-Email service, but the problem with that is that you can't follow private accounts/see followers-only toots. It would be great to have a full email bridge.

I was considering making this myself at one point. But I think one of the big problem with ActivityPub is that it describes a single particular account. So if my ActivityPub-email bridge was running you wouldn't also be able to access a Mastodon UI and for example browse other posts. So my account would need to be email-only which would be missing UX for a lot of things (like commenting on a random post I was linked to).

[–] kevincox@lemmy.ml 7 points 1 year ago

Firstly, Epic Games doesn't officially support Linux. So if anything goes wrong you may be out of luck.

However I have had success playing a handful of games on Lutris under Wine. I would check online sites for reviews of the games that you are most interested in and see how they worked for others. Wine is getting better every day so this will keep improving but I would at least look at the games you currently play and see what is up.

One major concern is games with DRM or anti-cheat. Most of these will not work on Wine, but providers are slowly adding support. But again, looking up the games you are interested in will tell you.

[–] kevincox@lemmy.ml 38 points 1 year ago (2 children)

This is pretty clever. As I understand it.

  1. Because LLMs are slow most of them stream the response to the user.
  2. The response is streamed as text, but generated in tokens.
  3. This means that each "chunk" leaks the length of the text corresponding to the token.
  4. You can then use heuristics to guess the text of the response based on the token lengths.

This is a good reminder any time you are sending content in small chunks over an encrypted channel, many encrypted channels don't provide protection against size leaks by default.

It seems there are a few easy solutions to this:

  1. Send the token IDs (as fixed-size integers) over the network rather than the text.
  2. Pad the text representations of the tokens to a fixed length.
  3. Batch the tokens more (and maybe add padding) to produce bigger chunks and obscure individual token size.

These still all leak the approximate length of the response, but that is probably acceptable.

[–] kevincox@lemmy.ml 11 points 1 year ago (1 children)

I hate this "costing people jobs". It is a sign of our broken society.

If this gate suitably replaces someone standing there we should implement it, raise taxes .000001% and put that money towards welfare/UBI. I'd rather just give people money than forcing them to do a useless job. Of course our society doesn't function nearly well enough for this to actually happen.

In this case the problem isn't the lost job but that this gate is an inconvenience for every shopper. They are wasting my time because it saves them .000001%.

[–] kevincox@lemmy.ml 1 points 1 year ago (1 children)

Yeah. Basically they can't stop you from leaving the store unless they have strong evidence that you committed a crime. Not showing a receipt is not strong evidence. Probably string evidence would be they saw you pick something up and followed you to the door without paying.

Apparently one exception may be Costco because they make you sign an agreement before entering. But I don't think this has ever been tried in court.

[–] kevincox@lemmy.ml 3 points 1 year ago (1 children)

Of course. My argument isn't that this is a good solution, I don't think it is. I just don't think safety is a notable criticism.

view more: ‹ prev next ›