Supabase is an open source Firebase alternative. Start your project with a Postgres database, Authentication, instant APIs, Edge Functions, Realtime subscriptions, and Storage.

Supabase

same, looks like there might be some sorta mixup somewhere

Can anyone point in my the right direction to learn about general postgres/relational database modeling techniques? I've tried looking online and have found some resources but they've been pretty shallow. Any help would be appreciated!

Supabase authentication is not working .. possible platform issue no response from <@&951828363943759883>

just there is some issues with authentication

and after one response from thread in help in discord there's no response in thread

**Free** tier using t4g.micro, what db instance type is used on **Pro** and **Enterprise**?

If you go to Database add-on plans here you can get an idea of the instance size options: https://supabase.com/pricing

For enterprise plan, it depends on your project. I would suggest a call with Rory. But anyone on Pro tier can get add-ons for upgraded instances:

Pro basically lifts the limits (database size, etc.) but doesn't change the hardware.  $5/mo gets you a t4g.small, which is an obvious upgrade, as it doubles the available RAM.

I missed that table on the pricing page. Thanks 👍 
Now I have to make some tests and convince my team to migrate from RDS

It was down last night for East and west US

hey. does supabase automatically get new access_tokens when they expire for third party providers? for example im using Spotify as a provider. When this token expires, will the supabase auth client get a new access token from Spotify?

No, I believe you need to manage that yourself and get a new key and change it in your Supabase dashboard.

I know the Apple token expires after 6 months and needs to be regenerated in this way.

storage is in the database right so could I have a foreign key being an image in storage and then have my own images table to add things like alt text, width, height and reference the image in storage and maybe even cascade delete it if the image is deleted

You can certainly link to the storage object table row, but you can't delete with cascade as the delete of the file data in s3 storage is done by the storage api which then deletes the object table row.  You would orphan the actual file in storage if you just delete the row in the object table.

I had to create a trigger function which called storage api thru http extension to delete the files "attached" to a note object when deleted in my app.

ok and for making policy what are all the fields I get? like I want to limit to certain image types for example 
```
((bucket_id = 'posts'::text) AND (storage.extension(name) = 'webp'::text))
```
but I can simply name the file test.webp and provide a jpg image like this but I want the actual file type not just name extension

when I list files I see  
```
metadata:
cacheControl: "max-age=3600"
mimetype: "image/webp"
size: 279338
```
perhaps somehow getting mimetype and size would beuseful to limit as well

You can get to those in RLS policies.  If you go to the dashboard table UI and select storage for the schema you can see the object table and columns you can use.  The metadata is stored in json.

Ok I am trying like this
```
(
  (bucket_id = 'posts'::text) AND 
  (
    ('private'::text = (storage.foldername(name))[1]) OR 
    ('public'::text = (storage.foldername(name))[1])
  ) AND 
  ((uid())::text = (storage.foldername(name))[2]) AND 
  (
    (
      (storage.extension(name) = 'webp'::text) AND
      (json_extract_path(storage.metadata, '{mimetype}') = 'image/webp'::text) AND
      (json_extract_path(storage.metadata, '{size}') < 999999 ) 
    ) OR 
    (storage.extension(name) = 'mp4'::text) OR 
    (storage.extension(name) = 'pdf'::text) OR 
    (storage.extension(name) = 'gif'::text)
  )
)(
  (bucket_id = 'posts'::text) AND 
  (
    ('private'::text = (storage.foldername(name))[1]) OR 
    ('public'::text = (storage.foldername(name))[1])
  ) AND 
  ((uid())::text = (storage.foldername(name))[2]) AND 
  (
    (
      (storage.extension(name) = 'webp'::text) AND
      (jsonb_extract_path(storage.metadata, '{mimetype}') = 'image/webp'::text) AND
      (jsonb_extract_path(storage.metadata, '{size}') < 999999 ) 
    ) OR 
    (storage.extension(name) = 'mp4'::text) OR 
    (storage.extension(name) = 'pdf'::text) OR 
    (storage.extension(name) = 'gif'::text)
  )
)
```
and get error on save `Error updating policy: missing FROM-clause entry for table "storage"Error updating policy: missing FROM-clause entry for table "storage"`
can I access the metadata like that or is this saying I need some sort of like from(storage) function

This is for an insert policy btw. And if I dont assign target roles I'm assuming just activates for all inserts

There is no function storage.metadata like foldername and extension, you have to do sql: 
select metadata from storage.objects where....
 and then do postgres json operators to get to your json field on that result.

area of focus being 
```
(json_extract_path(storage.metadata, '{mimetype}') = 'image/webp'::text)
```

i can do declare and begin and end in these checks then im assuming. Ill try that

its on storage.objects. When someone uploads file. I made it so there is posts bucket not public and then public folder and private folder and next folder in each is uid of uploader and want public ones public read for all and then all insert,update,delete for user on own files. Then only allow specified file types under specified size

Ahh, so there is likely a problem then...

The storage API does two calls to the object table when a file is uploaded.  First is an insert with basic information and to check policy.  Then AFTER it stores the file to s3 it updates and adds metadata....

ohh ok so there is really no way to check what is being uploaded or its size until after its too late. That's odd. In general is this a good approach I am doing

Yeah, so I think you just have the name of the file at insert time.  I can't remember what storage-api does if the 2nd update fails to the db.

https://github.com/supabase/supabase/discussions/6540

Ok I'll look into that. also so this is the read policy i have for public folder which means i can use getPublicUrl to get urls right? I seem to be having issue. and createSignedUrl() for private
```
supabase
        .storage
        .from('posts')
        .list('public/'+user.id, {
          limit: 100,
          offset: 0,
          sortBy: { column: 'name', order: 'asc' },
        }),
```
gets me the list of files and then i need urls from existing items
```
publicData.map(async (publicFile) => {
          const { publicURL, error } = await supabase
            .storage
            .from('posts')
            .getPublicUrl(`${user.id}/${publicFile.name}`, 60)
```
And it gives me urls but they dont work https://cnxmxunyrouuwprrtvst.supabase.co/storage/v1/object/public/posts/e8527e42-34de-4840-a378-c35f93a10570/test1.webp

getPublicUrl does not make an API call.  It just builds a path based on constants and parameters.  Then when you use that URL it does not get impacted by RLS policies at all as it goes straight to s3 storage.

To make a bucket public you need to actually select the "make public" option on the bucket.

oh so I guess instead of having  private posts bucket with public and private folders i should instead do public and private buckets with folders being uid/filename

Yes, if you want to have long lived urls without token then you need the bucket public.

ok I see what you mean. I have working properly just using token as is even for signed out users but may be advantages to switching to public anyway. 

Im guessing with public and as you referred to them "long lived urls" these might be better for seo because I am guessing tokenized way wouldnt show up in google images for example

Idk all the pros and cons of each but might give it a bit of thinking tonight

If I have say a tweet and its a 'draft' I would want that image private until send then could use moveto to make public on send. But I dont believe I can move from one bucket to another so would just need to make sure when they have a draft the img url is still public somewhere even if not convienent in front end

Move only works within a bucket.  If your urls are not predictable, then until you actually post the url somewhere public, the file is pretty secure as you can't list a public bucket without permission.  Of course depends on security level you need for unpublished info.

Hello, If I'm going to have multiple small projects (more than 2 projects, hobby project), am I going to be billed $25 per project? 

any way to get something ~$5/project (with more limited capability)?

I want to make an app which involves uploading files onto a server
But I don't want users to spam files which are duplicates / not relevant (NSFW for example)
What is the best way to manage this, is there a particular framework? And does the processing happen on device or serverside?

Hello! I've been reading through some of the github discussions and discord threads, and was wondering if there was any update on if supabase is SOC2 Type I/Type II compliant yet? The most recent discussion I found stated a Q1 2022 Target for SOC2 Type 1 compliance, has that been achieved? Is there any progress on others such as HIPAA?

Hello <@342679111350616066>, it's maybe the 8th time my supabase DB is down for like a month in prod - all my metrics are okay :

Really annoying, usually we find out and click on 'restart' pretty quickly, but now it's not even restarting for like 20 minutes.

We've replied via your support ticket and you can follow up there.  Thanks.

You need to (if you have not already) get support involved.  No one here is going to be able to help unless you have some errors to report.  Have you looked in the database postgres log to see if there is an error?

Nothing is replying on the infra, I can't access logs

sorry to hear about your experience Sacha. to second what Gary said, our support team is the best place to get help for this kind of issue. if you haven’t checked https://supabase.com/docs/support already that’s the best place to start