| 6 Mar 2024 |
 flokli | I don't think workers will be free if we send NAR files through them. | 12:44:42 |
| flokli | Someone needs to do the math, ensure the proposal is feasible, and the sponsors we do come up with can commit a bit long time | 12:45:06 |
 Julien | In reply to @domenkozar:matrix.org
it used to be one company paying for all of this Well that is one topic where I fully agree with you, but we also must be resilient to situations where we don’t have that solution working. The situation we currently face with GC/AWS cost shows that we are not ready to sustain the costs of hosting the cache with a cloud provider, and we do not own our own data | 12:45:16 |
| flokli | In reply to @domenkozar:matrix.org
we shouldn't let sunk cost fallacy destroy our history if we can do another plan of saving it We don't talk about destroying it, the plan was to put it into glacier, where we can get it back, it's just not immediately accessible, but much cheaper to store. | 12:45:43 |
 @domenkozar:matrix.org | we own the data, you can transfer it out any time | 12:45:47 |
| flokli | * We don't talk about destroying it, the plan was to put it into glacier, where we can get it back, it's just not immediately accessible, but much cheaper to store. Until we have a better place to host it long-term. | 12:46:01 |
| Julien | In reply to @domenkozar:matrix.org
we own the data, you can transfer it out any time But because we don’t have our own hardware this is kind of a dream | 12:46:14 |
| @domenkozar:matrix.org | In reply to @flokli:matrix.org
We don't talk about destroying it, the plan was to put it into glacier, where we can get it back, it's just not immediately accessible, but much cheaper to store. Until we have a better place to host it long-term. well that effectively makes it unusuable | 12:46:14 |
| @domenkozar:matrix.org | I feel like there are some really weird requirements coming up, when did we decide we need to own data? | 12:46:41 |
| Julien | In reply to @domenkozar:matrix.org
well that effectively makes it unusuable If we save it to glacier we can in the future recover it all at once on another infra ? | 12:46:43 |
| @domenkozar:matrix.org | we don't own github as well | 12:46:45 |
| @domenkozar:matrix.org | these are some really weird decisions to me | 12:47:22 |
| Julien | I'm not talking on behalf of anyone here, these are just my political biases | 12:47:49 |
| @domenkozar:matrix.org | my political bias is to make Nix succeed, at all cost | 12:48:08 |
| @domenkozar:matrix.org | this is not time nor place to bring our the morals because we all have our own and it's going to be impossible to find a common ground, that's how morals work | 12:48:55 |
| Julien | But I feel that currently we are obligated to delete some data we would not want to delete ideally just because we are not in control of the data. For example if we were, the scenario where we put in in CA storage would be more believable | 12:50:10 |
 Wanja Hentze | "it's fine to do everything via cloud" is as much of a moral position as "it's not fine to do everything via cloud" | 12:50:11 |
| flokli | I'm just saying we put a lot of effort into figuring out what data is pretty much unused, and to scrape the dataset. While there was a Opencollective open where companies could have sponsored a lot. We had one big contribution, but nothing to keep paying 7k or more a month for a sustained amount of time | 12:50:27 |
| flokli | Changing plans 20 times is very tiring | 12:50:41 |
| Julien | In reply to @julienmalka:matrix.org
But I feel that currently we are obligated to delete some data we would not want to delete ideally just because we are not in control of the data. For example if we were, the scenario where we put in in CA storage would be more believable That's what I mean by "we don't own the data". We have to obey arbitrary rules to do what we want with it | 12:50:46 |
| flokli | I'm not against changing the plan if it makes sense, but I'm not sure all the calculations are done to determine this solution is better, and we're in a position to execute it. | 12:51:37 |
| flokli | I'm pretty much burnt out of this whole matter. We announced an upcoming GC more than a month ago, I didn't see any sponsors step up to pay the bill so we don't have to GC. | 12:52:23 |
| @domenkozar:matrix.org | I can do my job and get that funding, but we need to agree first that we have trust established | 12:52:30 |
| Julien | In reply to @domenkozar:matrix.org
we don't own github as well well to get the code I can just git clone and some people are actually working towards exporting github PRs/issues history, it's the same question of trying not to be dependent on a vendor | 12:52:54 |
| @domenkozar:matrix.org | I have talked to companies and they are willing to sponsor, it just takes effort to make this thing happen | 12:53:00 |
| flokli | It's not my call on whether we can delay the GC further to evaluate another solution, that's a foundation topic. | 12:53:11 |
| flokli | So please reach out to them, and get them to make that call. | 12:53:20 |
| @domenkozar:matrix.org | alright I'll talk to them next week | 12:53:31 |
| @domenkozar:matrix.org | thanks guys :) | 12:53:35 |
| Julien | Anyway, in the short term any solution towards paying less and/or saving historical data is good | 12:53:59 |
