| 4 Aug 2022 |
 Minion3665 | Hmmm, well perhaps we should at least provide a pros/cons section for each of them rather than just some, as that way it's easier to compare
We could maybe even put them in a table but that's probably not necessary | 12:41:49 |
 Solène (she/her) | I think pros / cons should be removed, and just add a sentence in the first paragraph to tell they are listed in an order of the most preferable to the least practical | 12:42:47 |
| Solène (she/her) | using pros/cons is like having 3 wishes at a djin but you know each one has a drawback :D | 12:44:05 |
| Minion3665 | Haha yes perhaps
Although in this case it is somewhat accurate, so we at least do need to tell the reader about the drawbacks of each approach | 12:46:21 |
 fricklerhandwerk | I agree (and proposed) to list pros/cons, because this is reference. Be honest about what is going on, let the readers judge on their own. Then use the tutorial layer (i.e. nix.dev) to pick the recommended way and link to the reference. | 12:50:07 |
 Sandro 🐧 | just use search.nixos.org | 13:04:04 |
 pennae | how does search.nixos.org get to the data it displays? (if it read docbook we'll have to somehow fix that for the md migration) | 13:12:59 |
 Jan Tojnar | pennae: it uses pypandoc | 15:42:19 |
| Jan Tojnar | well, used before the rust migration | 15:42:38 |
| Jan Tojnar | https://github.com/NixOS/nixos-search/blob/c43ed8c85f11b041db2624cc249f3f1fb68760b2/flake-info/src/data/export.rs#L310 | 15:43:58 |
| pennae | if it can natively read MD there should be no extra trouble, right? | 16:16:38 |
| 5 Aug 2022 |
|  Kira Bruneau joined the room. | 01:24:48 |
| Kira Bruneau changed their display name from daxvena to Kira Bruneau. | 01:24:52 |
| 6 Aug 2022 |
 John Ericson | FYI to all, fricklerhandwerk's and I's reference (and some explanation docs were merged but commented out), and then I fixed up some conflicts and added more new material on my branch, opening https://github.com/NixOS/nix/pull/6877/files | 03:39:40 |
| John Ericson | I am curious what people think about this split format I went with, also perhaps of commenting out the material we can do some sort of build time flag to show/hide unstable docs | 03:41:01 |
| John Ericson | I do want users to see the new docs, but I also on fond of just brain dumping all the information and then stepping back to edit it into the right format | 03:41:52 |
 denna | In reply to @Ericson2314:matrix.org
I am curious what people think about this split format I went with, also perhaps of commenting out the material we can do some sort of build time flag to show/hide unstable docs "split format" means 1 sencence per line? yes i think that was fricklerhandwerk suggestion before and i think it is best for collaboratively written documentation.
Commented out "braindumps" are ok i think but there should be additional information so that these comments can also be removed if they are abandoned, information like "[2022.08.01 PR in preparation]" . | 15:23:39 |
| John Ericson | denna: oh the spit format I mean abstract vs concrete | 15:24:04 |
| John Ericson | I introduce a sort of "platonic nix" | 15:24:11 |
| John Ericson | and then I get into the gory details of files, the exact way we hash store objects, etc. | 15:24:26 |
| John Ericson | * and then I get into the gory details of files directories symlinks, the exact way we hash store objects and render store paths, etc. | 15:24:45 |
| denna | as a initial comment. i am suspicious of the suggestion to use pseudo code but need to render it to see how it feels. | 16:05:47 |
 tpw_rules | maybe it's me, but i'm not sure i like starting out the docs even farther away from actual nix | 16:30:52 |
| tpw_rules | oh i see there is commentary on this in the issue | 16:32:19 |
|  yuu[m] changed their display name from yuu to YourNick. | 17:41:38 |
| yuu[m] changed their display name from YourNick to yuu[m]. | 17:42:23 |
| John Ericson | denna: I am fine if the pseuco code data structures become diagrams some how | 17:58:33 |
| John Ericson | tpw_rules: FWIW my long term hope is that if Nix starts supporting things like WASI/WASM and whatever else, the abstract model will become more than idea in our heads, but also what ties everything together | 17:59:21 |
| John Ericson | I created new types of content addressing for the IPFS stuff for example (an option for "deep content addressing, were references are not arbitrary store paths but something more structured that prooves it is content addressed all the way down) | 18:00:26 |
| John Ericson | One could just document every instance of the general idea, but one is left then with a big bag of special cases with the underlying theme left as an exercise to the reader | 18:01:19 |
