🦋 Welcome to the MAIN() IRC channel of the Raku Programming Language (raku.org). Log available at irclogs.raku.org/raku/live.html . If you're a beginner, you can also check out the #raku-beginner channel!
Set by lizmat on 6 September 2022.
Voldenet my thoughts github.com/FCO/Red/issues/15#issue...1984826416 00:18
I'm not sure on `Red-*` methods, it's just an idea 00:19
SmokeMachine looking, thanks!
Voldenet created-on could use unix timestamp too
in fact, I could add that as a comment too 00:31
00:36 jpn joined 00:41 jpn left 01:25 Summer left, Summer joined
Voldenet defending that `keep model changes instead of model snapshot` approach is going to be hard to defend 01:28
ngl, it sounds cool, but nothing else 01:31
SmokeMachine Voldenet: I have added there an option to your migration "class"... github.com/FCO/Red/issues/15#issue...1984905771 01:49
Voldenet > User.^migration: *.drop-column("type"); 01:56
doesn't this require `User` entity to exist?
what if it gets removed in the future?
aruniecrisps @SmokeMachine you're suggesting that we use a .sqlite file to house the migration plan right? 02:07
SmokeMachine That's a reason I'm thinking in not only store the model representation but all model classes to each version
aruniecrisps: not the plan, but the history... 02:08
aruniecrisps: but after Voldenet's comment, I'm rethinking it...
aruniecrisps Yea didn't voldenet suggest using git diffs? 02:09
Voldenet well, depends
if we store snapshot, you can use diffs to extract differences
if we store differences in migrations, we have diffs and we reconstruct snapshots 02:10
though diffs are not useful in terms of frameworks
tbh. entity framework simply stores whole model in every migration 02:12
it gets beyond gigantic
so I'm sure what to avoid 02:13
aruniecrisps What do you think? 02:14
I wonder what Rails and Phoenix do
SmokeMachine I think my preferred way is storing all models for each version... that was we can get the difference between models and we are able to use them on migrations...
Voldenet that could work 02:15
SmokeMachine by diff, I don't mean git diff, but something like: Model1.diff: Model2
something like this: github.com/FCO/Red/blob/3e318994d1...akumod#L68 02:16
(this example uses diff from DB to model...)
aruniecrisps I'm assuming for that though you would need something like sqlite to store thee full model 02:17
SmokeMachine no, file
just copy the model file 02:18
aruniecrisps Doesn't that lead to what voldenet was talking about?
SmokeMachine what? 02:19
aruniecrisps Where if you store the whole model in every migration it gets way out of hand?
Or am I misunderstanding something
Voldenet Hmm, what if entities were only stored on change and otherwise old versions would be used 02:20
SmokeMachine sorry... I didn't get it...
Voldenet the project gets large if you maintain it long enough
SmokeMachine Voldenet: 02:21
> Hmm, what if entities were only stored on change and otherwise old versions would be used
I was thinking about that too...
Voldenet it would require some per-entity versioning and having the list of entities used in model 02:22
sounds horrible, but shouldn't be that tough to maintain
aruniecrisps what would that look like though?
what do you mean by per-entity versioning 02:23
SmokeMachine we just increment the version of a model if it changed on that migration...
s/just/only/
Voldenet `/migrations/entities/User/1/entity.raku: model User is rw { … }` 02:24
and then in the migration `/migrations/1/initial-migration.raku: use User:ver<1>` or something similar 02:25
aruniecrisps ah okay that makes sense
I'm not familiar enough with extended identifiers in Raku to figure how to pull out the version number if needed 02:26
SmokeMachine maybe we could use a custom CompUnit::Repository... 02:27
02:28 kylese left, kylese joined
SmokeMachine or maybe just add all of them on META6.json? 02:28
aruniecrisps I don't like the second idea
SmokeMachine why? 02:29
Voldenet project shouldn't be referencing other versions
aruniecrisps wait do you mean by add all of them to META6.json, what field would they go under
what voldenet said, and i feel that is the ability to get unwieldy very quickly 02:30
SmokeMachine provides
maybe the custom CompUnit::Repository then...
Voldenet consider this: `use lib <entities/User/1 entities/Roles/3 entities/Scores/2>; use User; use Roles…` 02:31
something like that sounds a bit awful
aruniecrisps Yea, that's what I'm thinking, I don't think the business logic should be directly referencing version numbers 02:32
Voldenet but it's a storage optimization for migrations then
SmokeMachine maybe return with this old idea: github.com/FCO/Red/issues/7 02:33
02:36 Summer left, Summer joined
aruniecrisps I think this idea might be better 02:37
What are your thoughts @Voldenet?
Voldenet sounds nice, the only limit is that it has to be parsed 02:38
with CompUnit::Repository only specific entities could need to be loaded 02:39
so, out of 100 entities in 4 versions, the latest necessary migration may only load 4 files (each 50b in size) instead of one being 20kb in size (I'm assuming 50b per entity) 02:40
it's not horrible though 02:41
aruniecrisps So basically @SmokeMachine if i'm getting you correctly the idea is to implement schema as a CompUnit::Repository? 02:45
SmokeMachine I'm thinking on something like: `use lib "RedSchema#User:ver<1>#Model2:ver<3>#Model3:ver<5>"`
aruniecrisps That feels like the same as use lib <entities/User/1 entities/Roles/3 entities/Scores/2> 02:48
SmokeMachine but you don't need to know where they are.. 02:49
but yes...
aruniecrisps you don't need to know where they are, but they're still referencing version numbers
SmokeMachine yes, and that's what we need on the migration, right? 02:50
Voldenet in normal code there wouldn't be any need for that, so either way it's managable 02:52
aruniecrisps @SmokeMachine oh you mean for the migration 02:53
i thought this use lib "RedSchema#User:ver<1>#Model2:ver<3>#Model3:ver<5> was for the business logic
SmokeMachine better, we could have a Schema.rakumod inside each migration dir and have: 02:55
class Schema:ver<2> does RedSchema { has @.models = ( Model1:ver<1>, Model2:ver<3>, Model3:ver<5>) }
and then, when we `use RedSchema` it would import all those Models...
I mean for migration, yes... but this last idea could also be used on normal code (with no versions...) 02:56
aruniecrisps I like this idea
SmokeMachine RedSchema would be just a simple `sub EXPORT` with the logic to se what models and what versions to load and just return a map with them... 02:57
(sorry, @.models would be strings... not the types) 02:58
the only problem with these ideas is to split files if there are more than 1 model on that... 02:59
3am... sorry I need to go now... 03:00
Voldenet productive night either way
4am here
aruniecrisps See you guys later! 03:01
SmokeMachine We need to add that last suggestion on the thread... 03:02
Voldenet I've added something 03:06
basically a very brief summary
03:06 Summer left 03:07 Summer joined
SmokeMachine thnks! I was doing that too... I replaced the class with a sub call (I think that makes more sense)... github.com/FCO/Red/issues/15#issue...1984964871 03:07
Voldenet: ^^
Voldenet and no thought was lost 03:09
SmokeMachine github.com/FCO/Red/issues/15#issue...1984966702
Thank you very much you 2!
I'm going to sleep
03:15 kylese left, kylese joined 03:37 Summer left 03:38 Summer joined 04:08 Summer left, Summer joined 04:39 Summer left, Summer joined 05:06 MasterDuke joined 05:09 Summer left 05:10 Summer joined 05:40 Summer left, Summer joined 06:11 Summer left, Summer joined 06:32 MasterDuke left 06:41 Summer left 06:42 Summer joined 07:11 kst left 07:12 Summer left, Summer joined 07:17 kst joined 07:34 jpn joined 07:39 jpn left 07:43 Summer left, Summer joined, Sgeo left 08:13 Summer left 08:14 Summer joined 08:44 El_Che left, Summer left, Summer joined 08:54 merp left 08:59 haxxelotto joined 09:01 merp joined, jpn joined 09:08 sena_kun joined 09:53 sena_kun left 10:42 ACfromTX left 11:13 hudo__ joined 11:17 hudo_ left 11:20 mark22k left 11:21 mark22k joined 11:23 mark22k left 11:24 mark22k joined 11:26 mark22k left 11:27 mark22k joined 11:33 haxxelotto left
dutchie what's the best way to make ascii-only regexes? 12:14
i.e. i don't want \w etc to match non-ascii characters. or would it be better just to check my string is ascii separately from whether it matches the regex 12:15
lizmat m: say "aa" ~~ / <:ascii>+ / 12:36
camelia ｢aa｣
lizmat m: say "bđź‘Ťa" ~~ / <:ascii>+ / 12:38
camelia ｢b｣
lizmat m: say "bđź‘Ťa" ~~ / ^ <:ascii>+ $ /
camelia Nil
lizmat m: say "ba" ~~ / ^ <:ascii>+ $ /
camelia ｢ba｣
lizmat m: say "bđź‘Ťa" ~~ / <-:ascii>+ / 12:39
camelia ｢👍｣
lizmat ^^ dutchie
12:53 MasterDuke joined 12:57 jpn left
dutchie ah :ascii, couldn't find that in the docs anywhere 12:59
lizmat m: say "bđź‘Ťa".contains( / <-:ascii> /' 13:20
camelia ===SORRY!=== Error while compiling <tmp>
Unable to parse expression in argument list; couldn't find final ')' (corresponding starter was at line 1)
at <tmp>:1
------> say "b👍a".contains( / <-:ascii> /⏏'
expecting any of:…
lizmat m: say "bđź‘Ťa".contains( / <-:ascii> /
camelia ===SORRY!=== Error while compiling <tmp>
Unable to parse expression in argument list; couldn't find final ')' (corresponding starter was at line 1)
at <tmp>:1
------> say "b👍a".contains( / <-:ascii> /⏏<EOL>
lizmat m: say "bđź‘Ťa".contains( / <-:ascii> / ) 13:21
camelia True
lizmat m: say "ba".contains( / <-:ascii> / )
camelia False
lizmat that's a quick way to ensure that a string contains any non-ascii grapheme
[Coke] dutchie - if you could be so kind as to open a raku/doc ticket about what you were trying to find. I just tried to search for :ascii itself and couldn't find it, but someone trying to answer your question wouldn't start with that. (but we should probably have all the regex :classes indexed for search.) 13:22
If not, no worries
tbrowder__ hi 13:32
hi 13:33
hi 13:35
lizmat tbrowder__ o/ 13:36
tbrowder__ hi, lizmat. i'm still blind to my msgs 13:37
i'm working on a module that looks for the user to set a certain value by one of three mean: 13:39
*means: config file, env var, or CLI. i usually expect the priority for programs to check in order first to last: cli, env var, config. 13:42
i would appreciate other opinions as to the best way to do that. maybe POSIX has an answer 13:43
13:45 jpn joined
tbrowder__ [Coke]: any problem with a merge of my File::Temp PR? 13:48
13:59 jpn left 14:13 jpn joined
[Coke] tbrowder__: I only had an issue with the sha items in the changelog, but am fine if they go in. 14:15
I can't address the "meat" of the changes. 14:16
tbrowder__ ok, thnx, maybe lizmat could take a look when time permits 14:22
14:22 haxxelotto joined 14:28 xinming left 14:29 xinming joined 15:02 jmcgnh left 15:35 haxxelotto_ joined, haxxelotto left 15:49 Sgeo joined 16:37 vlad joined 16:48 vlad left, vlad joined 17:05 El_Che joined
_grenzo From a US/Pacific time point of view, rakoons (raccoons?) are crepuscular 17:08
17:08 vlad left, vlad joined
jdv trash pandas? 17:10
_grenzo Sure, as long as they use Raku 17:11
17:51 haxxelotto_ left, haxxelotto joined 18:11 haxxelotto left 18:21 vlad left 18:47 jpn left
antononcube @grenzo Visual aid: 19:12
cdn.discordapp.com/attachments/633...80191&
tbrowder__ hi 19:30
humbug! 19:31
19:34 jpn joined
_grenzo Curious, do images work across the irc bridge? 19:34
19:38 jpn left 19:39 haxxelotto joined 20:01 abraxxa-home joined 20:02 abraxxa-home left 20:04 abraxxa-home joined 20:27 ericst joined 20:28 jpn joined, ericst left 20:32 kst left, jpn left 20:39 kst joined 20:43 jpn joined 20:48 jpn left 20:57 abraxxa-home left 21:23 jpn joined 21:28 jpn left
japhb _grenzo: We get a link to a WebP file on cdn.discordapp.com 21:42
21:44 jpn joined 21:46 haxxelotto left 21:48 jpn left 22:00 haxxelotto joined 22:08 sena_kun joined 22:11 jmcgnh joined 22:12 MasterDuke left 22:24 ponycat joined
ponycat does raku have an equivalent to this? docs.python.org/3/library/os.path.....path.join 22:26
antononcube @ponycat Yes, docs.raku.org/type/IO/Path 22:29
ponycat thanks, it looks like the add method is what I was looking for in particular 22:30
aruniecrisps @SmokeMachine does .^save return the id of the new record when you're saving a new record? 22:33
SmokeMachine No, but you can .^refresh to repopulate your object (if you are using the refresh experimental) 22:35
(If I remember it well… let me take a look) 22:36
Sorry, I was wrong, .^save DOES return stuff: github.com/FCO/Red/blob/master/lib...kumod#L481 22:42
22:44 jpn joined 22:45 MasterDuke joined
aruniecrisps It does return something, question is what does it return :cameliathink: 22:46
22:49 jpn left
SmokeMachine In Pg we return * 22:55
For SQLite, I think it doesn’t return anything useful… but I have no access to a computer to be sure now… 22:57
But why are you using .^save instead of .^create?
aruniecrisps @SmokeMachine I'm thinking about a good way of doing validations 22:58
SmokeMachine What kind of validation? 22:59
aruniecrisps form data validation, checking to see whether or not the form data is good or not
so say i'm trying to check to see whether or not a value that is entered into a field is valid 23:00
the way i have it is I just validate this data directly and accumulate the errors to show to the client
if you look at what i'm doing for begumpura you'll see what i mean
SmokeMachine You could do validations on type `subset Name of Str where 0 < *.chars <= 30; has Name $.name is column{ :type<varchar(30) };` 23:04
But I think form validations should be done before data gets to the model… 23:05
aruniecrisps That's probably a good call
SmokeMachine As I do on this HTML::Component example: github.com/FCO/HTML-Component/blob...akumod#L32 23:07
And that’s common to a form validation be different to the model validation… 23:08
23:08 haxxelotto left 23:09 haxxelotto joined
SmokeMachine aruniecrisps: make sense? 23:19
23:36 sena_kun left 23:45 jpn joined 23:51 jpn left

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!