janglad

124 posts

janglad

janglad

@_janglad

jr product engineer @lemni, sometimes keyboard stuff @ https://t.co/bdQa3DyvwY

België Katılım Mayıs 2013
394 Takip Edilen26 Takipçiler
janglad
janglad@_janglad·
@AdamRackis depending on the type of app I’ve come around to it. “One mistake” is true for app code too, and RLS gives you a centralised place you can audit. Requires you to basically put a team/user id on every table tho and gets painful once you touch roles.
English
0
0
0
379
Adam Rackis
Adam Rackis@AdamRackis·
My god love seeing a well-respected person share this view with me
David Cramer@zeeg

RLS was a mistake and folks exposing that level of complexity to less technical users is asking for trouble. It was a mistake in Firebase. It’s a mistake in Supabase. It will be a mistake in the next product too. I personally - even knowing how to secure it - would never touch it. It’s the worst security footgun you can imagine. One small mistake and your data is available to the world.

English
7
2
144
24K
Krijn
Krijn@krijnrijshouwer·
Finished 33rd in what is often called the toughest footrace on earth. The 40th Marathon Des Sables. 270 km through the Sahara. 1,500 starters. It was brutally hot, deeply painful, and profoundly magical. Gave the desert everything 🏜️
Krijn tweet media
English
22
0
61
3.5K
janglad
janglad@_janglad·
@steveruizok This is not what I imagined when they said AI is going to turn us all into PMs
English
0
0
7
547
Steve Ruiz
Steve Ruiz@steveruizok·
thank you JavaScript
Steve Ruiz tweet media
English
16
4
453
39K
janglad
janglad@_janglad·
@mattpocockuk Por que no los dos? W/ cursor I like doing a heavy grill me up front to outline all general requirements, implement in smaller phases that get a quick plan made for them. Grill me again if something core changes.
English
0
0
0
107
Matt Pocock
Matt Pocock@mattpocockuk·
There was some confusion around this, so let me clarify. - I don't use plan mode - I still plan like crazy, using my skills /grill-me, /write-a-prd, then /prd-to-issues - Bad plans = Bad outputs
Matt Pocock@mattpocockuk

I have also stopped using plan mode It creates a plan FAR too eagerly and usually asks you zero questions en route The whole point of planning is to get on the same wavelength with the LLM, not to generate an asset you don't read /grill-me all the way

English
71
34
683
74.2K
janglad
janglad@_janglad·
@saltyAom Wouldn’t call that a nit 🤷
English
0
0
0
237
SaltyAom
SaltyAom@saltyAom·
From my experience, a lot of people I talk to really like Bun But most nitpick problems are usually memory usage/leak
English
13
1
118
9.6K
janglad
janglad@_janglad·
@steveruizok Can I also resize 19000 at once? This would be a big unlock for my workflow
English
1
0
15
41.1K
Steve Ruiz
Steve Ruiz@steveruizok·
selecting and resizing 18000 unique shapes (do not try at home)
English
7
1
49
5.2K
janglad
janglad@_janglad·
@__morse I can see the benefits and while this could be solved I do like that something so prone to malicious prompt injection is local.
English
1
0
2
785
Tommy D. Rossi
Tommy D. Rossi@__morse·
skills should have been urls not local files. brain dead spec
English
15
0
57
14.7K
janglad
janglad@_janglad·
@cramforce However I assume Doc could be inferred based on the type of kv + “docs/“?
English
0
0
0
64
janglad
janglad@_janglad·
@cramforce If you’re after partial inference where you provide only 1 type param, that’s not possible sadly
English
1
0
1
1K
janglad
janglad@_janglad·
@prisma I am aware of that, but all of these are way less ergonomic. Either you query in your .ts file without any type safety, or you used TypedSQL but that introduces custom limits/rules about arguments, file naming, organisation, ... .
English
1
0
0
24
Prisma Postgres
Prisma Postgres@prisma·
Prisma 7 AMA is open. Curious about usage, performance, or what’s changed under the hood? Drop your questions in the 🧵 We’ll collect them, get answers from our engineers & publish a blog with the highlights.
Prisma Postgres tweet media
English
8
4
16
3.1K
janglad
janglad@_janglad·
@prisma I mean something similar to Drizzle's sql`` helper that you can incorporate into TS queries instead of dropping down to SQL completely.
English
1
0
0
35
janglad
janglad@_janglad·
@prisma Are there any future plans on making it easier to use snippets of SQL? e.g. to be able to use postgis, write a more complex where clause, ... . Or is this completely incompatible with current internals. Typed SQL is nice but has its own sets of drawbacks.
English
1
0
0
109
janglad
janglad@_janglad·
@scottjla @benhylak I mean what does that have to do with RLS? It's a standard PG feature.
English
0
0
1
14
Scott
Scott@scottjla·
@benhylak Part of the trick with every cloud service is figuring out their trick on how they lock people in and then start over charging them for services that are available at a fraction of the cost elsewhere
English
1
0
4
621
ben hylak
ben hylak@benhylak·
we use supabase, and like supabase, but i will say that theo has a very good point here re: RLS. have seen so many people get this wrong. makes it too easy to make big mistakes.
Theo - t3.gg@theo

More fun things I just remembered: - On Supabase, if you have ONE project that goes over free limits, you have to move to a "pro" account, which charges you $10 PER PROJECT - Supabase's patterns of "connecting to the db" put you in RLS hell, which is HORRIBLE for both devs and agents alike - Branching for dev and PRs in Supabase is rough (I don't even think about this in Convex) There's so many more things I'm sure I'm forgetting. Here's a video I did breaking it all down (that I think may have triggered the current Supabase cringe posting) youtube.com/watch?v=B6C-MW…

English
7
0
72
25.7K
David K 🎹
David K 🎹@DavidKPiano·
@IroncladDev You can, I'm not your boss But this is nice for reusable components and preventing ID conflicts and/or hydration errors if you use random string IDs
English
3
0
60
3.9K
David K 🎹
David K 🎹@DavidKPiano·
Quick React reminder that `useId()` is underrated You can `useId(…)` + the `form={…}` attribute to submit forms from buttons outside the form element Avoids the whole `useEffect()` or `useRef()` juggling you'd otherwise have to do
David K 🎹 tweet media
English
42
87
1.2K
118.4K
janglad
janglad@_janglad·
@codinginflow #opting-out-specific-components" target="_blank" rel="nofollow noopener">react.dev/learn/react-co…
QME
0
0
1
24
Coding in Flow
Coding in Flow@codinginflow·
I can't figure out this React problem: The FormUnsavedChangesGuard prevents accidental form closes for dirty forms. Problem: When I try to navigate away after a successful mutation, the form state is still dirty because `useFormState` updates on the next render.
Coding in Flow tweet media
English
5
1
6
715
janglad
janglad@_janglad·
@harrysolovay If you're looking to get somewhat typed decoding have a look at Drizzleify here #L621" target="_blank" rel="nofollow noopener">github.com/janglad/drizzl… Bit jank but turns e.g. {a: string, b: string} | {a: string; c:string} into {a: string, b?: string, c?: string}.
English
0
0
1
19
Harry Solovay
Harry Solovay@harrysolovay·
How do you guys approach working with tagged union values with drizzle? Any techniques that made this easier for you?
English
3
0
3
476
janglad
janglad@_janglad·
@DrizzleORM @colinhacks 'compiling' the QB -> SQL step is basically what .prepare() already does under the hood right?
English
0
0
0
44
janglad
janglad@_janglad·
@DrizzleORM @colinhacks Out of interest, what would be the max perf gained here? I'd guess there's not _that_ much work done when going from QB -> SQL. Do they (or RQB) queries "scale" poorly when they get huge? Or would focus more me on aggregating data for RQB queries?
English
3
0
0
194
Drizzle ORM
Drizzle ORM@DrizzleORM·
Anyone has experience with JIT functions? Any caveats to look for apart from different runtimes not supporting new Function? Perf boosts are insane, we would love to make sure there are no surprises down the road
English
11
0
71
13.7K

Keşfet

@AdamRackis @maxvoltar @krijnrijshouwer @noudadrichem @steveruizok @mattpocockuk @saltyAom @__morse