Continuation of prisma/prisma#593

Env vars are now disallowed in the provider argument of datasources

datasource ds {
  provider = env("DB_PROVIDER")
  url = env("DB_URL")
}

This should also be done in the VS Code extension.

Is there a way to automatically test new builds of this in VSCode somehow?

works on windows vscode well.
but I moved project to ubuntu and installed same plugin prisma extension.

but all type and fileds shows error even if it is correct.

When having inline strings of GraphQL like this:

const gql = /* GraphQL */`
  query {
    isHealthy
  }
`

The VSCode GraphQL extensions enable syntax highlighting for that multiline string.

It would be great to have the same for a Prisma schema:

const schema = /* Prisma */`
  model User {
    id Int @id 
  }
`

Steps to reproduction:

1. Copy example from into a new file https://github.com/prisma/prisma2/blob/master/docs/data-modeling.md#example
2. Changed one of the data types to something that does not exist. No red squiggly line.
3. Added some spacing between field name and type - then saved. Nothing changed.

(Windows, Extension version 0.0.24, VSCode 1.42.1)

Currently the @@id block attribute for compound ids does not seem to be supported.

type User {
  id: ID! @id
  name: String!
}

The only change it made to the above is it made @id yellow. The rest of the text was unchanged.

Right now, many Prisma1 users accidentally install this extension and then get a lot of errors because the content of the file of course is not valid.

The solution to their problem is to install https://marketplace.visualstudio.com/items?itemName=Prisma.vscode-graphql and change their file extension to .graphql so the new extension does not get triggered by default.

Maybe we can recognize GraphQL files and instead of just underlining everything as invalid, give the user some feedback what is going on and point them in the right direction?

Cannot install prisma-fmt: prisma-fmt: Unsupported OS platform: win32

When modeling your schema, dealing with relations is a very common workflow. There are multiple ways how the Prisma VSC extension could be helpful:

• Code action to add missing back-relation
• Linting + code action when a relation is ambiguous
  • When there are multiple relations between the same types and/or self-relations
  • In the case of a 1:1 relation and the referencing key is not specified
• Show implicit/default values

We should also make sure to bump the plugin version on each release to make sure that users always use a recent version of the prisma-fmt binary.

via prisma/prisma-engine#212 (comment)

via #31 (comment)

As we have releases of this out there, we should also have release here on GitHub.

Using the same env var for provider does not work:

Details:

Even if using env vars is not allowed here, or using it twice does not make sense, the error message should be better.

Can we please have the option to disable vertical alignment in the Auto-formatter. Thanks :)

model User {
    id Int @id
    enum TestEnum
}

enum TestEnum{
  A
  B
}

There is no color highlighting for the Enum.

Be able to click on relation, goes to the correct model.

How does it work?
What does it look like?

Hi. I had comments in my schema file which use to be fine a month ago but now, these come up as errors. Screenshot below. It shows it as unexpected end of block and if I remove the comment, the error goes away. Feels like a bug. Or am I doing anything wrong?

To reproduce introspect this SQL schema with the latest alpha. prisma/prisma#1599

This issue is related to the auto-format feature, which formats the .prisma file every time I press save in VS Code.

Previously an enum would be auto-formatted so that each option is one a separate line (see "expected behavior" below) but lately it seems to format it so that all options are on one line (see "actual behavior" below).

Even when I manually format it the expected way, when I press save the auto-formatter will change all my enums to the other format.

[Expected Behavior]

[Actual Behavior]

I'm using version 0.0.23 of the plugin, with latest version of VS Code at the time of writing.

When formatting,

id Int @id // This is the ID

becomes

id Int @id
// This is the ID

The comment should stay at the end of the line.

Proposed solution

Modify prisma-fmt grammar: To the end of field delarations (and similar), add - SPACE* - COMMENT? - NEWLINE?, where - is the non-trivia concatenation operator as discussed here.

Then, we can safely distinguish between comments in the same line (i.e. belongs to the field) and other comments when formatting.

When you already have entered a select or include statement in Prisma Client, you can't see the JSDoc anymore without scrolling, as it gets pushed down.

I created this issue in this repo, because this is something we could fix in our VSCode extension.

This is more a feature request than a bug
Right now the Extension says:

This plugin is designed for Prisma 2. Information about the new datamodel syntax can be found....

But i think it would be awesome to have some rc file to specify the Prisma version so that the extension will load the proper Syntax and formatting, and if no one is specified, take the last one.

I think it would be great for the community and also is scalable for future major releases

Is there a way to use the alpha version of this extension, that has the same validation as the current prisma2@alpha?

It would be nice if you could visually differentiate scalar from non scalar fields, as they are quite different.

Ideas:

• Slightly different text color
• Maybe we can render a an icon next to it? Something like the link icon here on GH maybe
  

Related: #45

With the latest preview change arrays of primitives utilise Postgres's native support for this feature.

The VSCode plugin however does not take this into account and displays an error:

Note I have specified Postgres:

datasource db {
  provider = "postgresql"
  url      = env("POSTGRES_URL")
}

The idea is to be able to Cmd + Click on the types in a .prisma file, to go to some information of what they actually mean (e.g. postgres.schema).

We should investigate how this could work.

Currently:
https://github.com/prisma/vscode/blob/23ad4d68150e3d97ef7756a2f8308556b21c5ee9/src/install.ts#L14-L17

This might not work in the future as we changed the domain to something else.

Big project requires to split models into multiple files. But sometimes models use other models in definition (you're right, for relations!).

Now:

Should be:

Import models or extension should look at prisma.yml and found related models:

Hi. I am not sure if I am doing something wrong, but since today, automatic formatting/alignment of schema.prisma file no longer works. It used to work properly before.

Noticed that there is some version upgrade. Any configuration to change from my end or is it a bug?

Open this file in VSCode and apply "Prisma" highlighting:

$ /Users/runner/runners/2.164.0/work/introspection-engine-output/introspection-engine-output/node_modules/.bin/prisma2 introspect --url mysql://prisma:qd58rcCywPRS4Stk@mysql127-divy.cg7tbvsdqlrs.eu-central-1.rds.amazonaws.com:3306/Basketball_men --print
datasource db {
  provider = "mysql"
  url      = "mysql://prisma:qd58rcCywPRS4Stk@mysql127-divy.cg7tbvsdqlrs.eu-central-1.rds.amazonaws.com:3306/Basketball_men"
}

model awards_coaches {
  award   String?
  id      Int      @default(autoincrement()) @id
  lgID    String?
  note    String?
  coaches coaches? @map(["coachID", "year"]) @relation(references: [coachID, year])

  @@index([coachID], name: "coachID")
  @@index([coaches], name: "coachID_2")
  @@index([coaches], name: "coachID_3")
  @@index([year, coachID], name: "year")
  @@index([year, coachID], name: "year_2")
  @@index([year, coachID, lgID], name: "year_3")
}

model awards_players {
  award    String
  lgID     String?
  note     String?
  pos      String?
  year     Int
  playerID players

  @@id([playerID, year, award])
  @@index([playerID], name: "playerID")
}

model coaches {
  coachID          String
  lgID             String?
  lost             Int?
  post_losses      Int?
  post_wins        Int?
  stint            Int
  won              Int?
  teams            teams            @map(["tmID", "year"])
  awards_coacheses awards_coaches[]

  @@id([coachID, year, tmID, stint])
  @@index([coachID], name: "coachID")
  @@index([teams], name: "tmID")
  @@index([teams], name: "tmID_2")
  @@index([year, coachID], name: "year")
  @@index([year, coachID, lgID], name: "year_2")
  @@index([year, tmID], name: "year_3")
  @@index([year, tmID], name: "year_4")
}

model draft {
  draftFrom      String?
  draftOverall   Int?
  draftRound     Int?
  draftSelection Int?
  firstName      String?
  id             Int     @default(0) @id
  lastName       String?
  lgID           String?
  playerID       String?
  suffixName     String?
  teams          teams?  @map(["tmID", "draftYear"])

  @@index([teams], name: "tmID")
}

model players {
  birthCity        String?
  birthCountry     String?
  birthDate        DateTime?
  birthState       String?
  college          String?
  collegeOther     String?
  deathDate        DateTime?
  firstName        String?
  firstseason      Int?
  fullGivenName    String?
  height           Float?
  highSchool       String?
  hsCity           String?
  hsCountry        String?
  hsState          String?
  lastName         String?
  lastseason       Int?
  middleName       String?
  nameGiven        String?
  nameNick         String?
  nameSuffix       String?
  playerID         String           @id
  pos              String?
  race             String?
  useFirst         String?
  weight           Int?
  awards_playerses awards_players[]
  players_teamses  players_teams[]  @relation(references: [playerID])
  player_allstars  player_allstar[]
}

model players_teams {
  assists            Int?
  blocks             Int?
  dRebounds          Int?
  fgAttempted        Int?
  fgMade             Int?
  ftAttempted        Int?
  ftMade             Int?
  GP                 Int?
  GS                 Int?
  id                 Int     @default(autoincrement()) @id
  lgID               String?
  minutes            Int?
  note               String?
  oRebounds          Int?
  PF                 Int?
  points             Int?
  PostAssists        Int?
  PostBlocks         Int?
  PostdRebounds      Int?
  PostfgAttempted    Int?
  PostfgMade         Int?
  PostftAttempted    Int?
  PostftMade         Int?
  PostGP             Int?
  PostGS             Int?
  PostMinutes        Int?
  PostoRebounds      Int?
  PostPF             Int?
  PostPoints         Int?
  PostRebounds       Int?
  PostSteals         Int?
  PostthreeAttempted Int?
  PostthreeMade      Int?
  PostTurnovers      Int?
  rebounds           Int?
  steals             Int?
  stint              Int?
  threeAttempted     Int?
  threeMade          Int?
  turnovers          Int?
  playerID           players
  teams              teams?  @map(["tmID", "year"])

  @@index([fgAttempted], name: "fgAttempted")
  @@index([fgAttempted], name: "fgAttempted_2")
  @@index([playerID], name: "playerID")
  @@index([teams], name: "tmID")
  @@index([teams], name: "tmID_2")
  @@index([teams], name: "tmID_3")
  @@index([year, tmID], name: "year")
}

model player_allstar {
  assists         Int?
  blocks          Int?
  conference      String?
  d_rebounds      Int?
  fg_attempted    Int?
  fg_made         Int?
  first_name      String?
  ft_attempted    Int?
  ft_made         Int?
  games_played    Int?
  last_name       String?
  league_id       String?
  minutes         Int?
  o_rebounds      Int?
  personal_fouls  Int?
  points          Int?
  rebounds        Int?
  season_id       Int
  steals          Int?
  three_attempted Int?
  three_made      Int?
  turnovers       Int?
  playerID        players

  @@id([playerID, season_id])
  @@index([playerID], name: "player_id")
}

model series_post {
  id         Int     @default(autoincrement()) @id
  L          Int?
  lgIDLoser  String?
  lgIDWinner String?
  round      String?
  series     String?
  W          Int?
  teams      teams?  @map(["tmIDLoser", "year"]) @relation("series_post_tmIDLoser_yearToteams")
  teams      teams?  @map(["tmIDWinner", "year"]) @relation("series_post_tmIDWinner_yearToteams")

  @@index([teams], name: "tmIDLoser")
  @@index([teams], name: "tmIDWinner")
}

model teams {
  arena                                           String?
  attendance                                      Int?
  awayLost                                        Int?
  awayWon                                         Int?
  bbtmID                                          String?
  confID                                          String?
  confLoss                                        Int?
  confRank                                        Int?
  confWon                                         Int?
  divID                                           String?
  divLoss                                         Int?
  divWon                                          Int?
  d_3pa                                           Int?
  d_3pm                                           Int?
  d_asts                                          Int?
  d_blk                                           Int?
  d_dreb                                          Int?
  d_fga                                           Int?
  d_fgm                                           Int?
  d_fta                                           Int?
  d_ftm                                           Int?
  d_oreb                                          Int?
  d_pf                                            Int?
  d_pts                                           Int?
  d_reb                                           Int?
  d_stl                                           Int?
  d_tmRebound                                     Int?
  d_to                                            Int?
  franchID                                        String?
  games                                           Int?
  homeLost                                        Int?
  homeWon                                         Int?
  lgID                                            String?
  lost                                            Int?
  min                                             Int?
  name                                            String?
  neutLoss                                        Int?
  neutWon                                         Int?
  o_3pa                                           Int?
  o_3pm                                           Int?
  o_asts                                          Int?
  o_blk                                           Int?
  o_dreb                                          Int?
  o_fga                                           Int?
  o_fgm                                           Int?
  o_fta                                           Int?
  o_ftm                                           Int?
  o_oreb                                          Int?
  o_pf                                            Int?
  o_pts                                           Int?
  o_reb                                           Int?
  o_stl                                           Int?
  o_tmRebound                                     Int?
  o_to                                            Int?
  pace                                            Int?
  playoff                                         String?
  rank                                            Int?
  tmID                                            String
  won                                             Int?
  year                                            Int
  coacheses                                       coaches[]
  drafts                                          draft[]
  players_teamses                                 players_teams[]
  series_posts_series_post_tmIDLoser_yearToteams  series_post[]   @relation("series_post_tmIDLoser_yearToteams")
  series_posts_series_post_tmIDWinner_yearToteams series_post[]   @relation("series_post_tmIDWinner_yearToteams")

  @@id([year, tmID])
  @@index([tmID], name: "tmID")
  @@index([tmID, year], name: "tmID_2")
}

Then remove the first line, to see the actual errors.

It would be nice if you could still see these errors even with an invalid first line.
(The invalidity of the first line is also super easy to muss unfortunately as only the first character is highlighted)

VSC indicates an error for the question mark in Post[]?, even though Prisma2 docs allows it.

model User {
  ...
  posts   Post[]?   @relation(name:"OneUserToManyOptionalPosts")
}

model Post {
  ...
  User   User!   @relation(name:"OneUserToManyOptionalPosts")
}

Here's a preview to our current syntax theme. I suggest we do some more fine-tuning e.g. related to how = is rendered.

Compare this to Terraform for example:

Hey guys,

as much as I am looking forward to using Prisma 2 and especially the new datamodel since it includes much needed features, I am a bit aghast about the way the transition is handled.

Let's be honest, Prisma 2 is not even a mature thing yet and production-readiness is also a thing for the future. Right now we have .prisma files in the v1 schema and are definitely stuck with them for a while, but since you are updating this plugin to match the v2 schema already, wouldn't it be nice to leave a v1-compatible version out here?

This is not working in the extension as it is using an outdated URL for download . The right URL can be found in the binaries spec .

We should also make sure to bump the plugin version on each release to make sure that users always use a recent version of the prisma-fmt binary.

via prisma/prisma-engines#212 (comment)

I like to separate out scalars and relations in a model definition, but then, the auto-formatter messes up the spacing.

Steps to reproduce:

In a model with two fields, id & name, I'd expect the formatted output to be:

model User {
  id    Int     @id

  name  String
}

But instead, the auto-formatter changes it to:

model User {
  id Int @id

  name String
}

Hi, I'm running on Mac OS 10.14.6 with prisma2 installed globally via yarn. When I install and run the vscode extension on a prisma2-formatted schema.prisma, I get this error in my vscode output:

[2019-08-14 11:51:05.192] [renderer1] [error] spawn /Users/nelsonpecora/.vscode/extensions/prisma.prisma-0.0.15/out/prisma-fmt EACCES: Error: spawn /Users/<user>/.vscode/extensions/prisma.prisma-0.0.15/out/prisma-fmt EACCES
	at Process.ChildProcess._handle.onexit (internal/child_process.js:232:19)
	at onErrorNT (internal/child_process.js:407:16)
	at process._tickCallback (internal/process/next_tick.js:63:19)

There are a number of reserved model names that can be found here: https://github.com/prisma/prisma-client-js/blob/master/packages/photon/src/generation/generateClient.ts#L342-L492

Rather than only documenting these reserved keywords, I think it would be helpful if this was already caught in the editor with an error and red squiggly lines.

.prisma files should get a custom icon. Please see this SO discussion on how.

The released version of this extension should pin to a version/hash of the binary that is also used in the release of the CLI.

Right now I believe we're formatting whenever the document has changed. We should probably hook into VSCode better to support:

1. Adherence to "Format on Save" option
2. Format document command

Hello,

I'm getting syntax errors for .prisma files. Could you please take a look? I am using windows 8.1.

Thanks,

Trung

type Link {
id: ID! @id
createdAt: DateTime! @createdat
description: String!
url: String!
}

above query works.

model Link {
id Int @id
createdAt DateTime
description String
url String
}

above query emit error

ERROR: Syntax error while parsing GraphQL query. Invalid input '/', expected OperationDefinition, FragmentDefinition or TypeSystemDefinition (line 1, column 1):
// type Link {
^

{
"data": {
"deploy": null
},
"errors": [
{
"locations": [
{
"line": 2,
"column": 9
}
],
"path": [
"deploy"
],
"code": 3017,
"message": "Syntax error while parsing GraphQL query. Invalid input '/', expected OperationDefinition, FragmentDefinition or TypeSystemDefinition (line 1, column 1):\n// type Link {\n^",
"requestId": "local:cjyq4fxd4001b0739xws5xc59"
}
],
"status": 200
}

Model declarations have to be indicated with the model keyword.
and when i change token to model
ofcource,
Syntax Error: Unexpected Name "model"

does not matter really, but kind of annoying

When trying to auto-format a schema.prisma file from a prisma2 starter project, VS Code 1.40.0 emits the following:

I have the extension installed, as seen below. Am I doing something wrong?

Edit: It turns out that Windows is not yet supported.

The GraphQL VSCode extension has a nice feature:
In JS code you can activate syntax highlighting of a multiline string like so:

const gql = /* GraphQL */ `
{
  users {
    id
  }
}
`

I would like to have that also for the Prisma definition language:

const gql = /* Prisma */ `
model User {
  id    Int @id
}
`

I think supporting both lowercase prisma and uppercase Prisma makes sense here.

Right now @timsuchanek has to manually ping @matthewmueller to execute a release of the VSCode extension. This is error prone and can be forgotten. We should automate that and integrate it into our release pipeline.