Best Of

LINQPad is licensed by user, so each user that wants to the product will need a license. It sounds like you might want to consider a team or Enterprise license in this scenario.

More work than I thought. Am prioritizing results to data grids ATM.

I'd like to request a feature enhancement to allow customization of the system prompt used in the "Trigger AI completion" action. Being able to define or override the system prompt would provide greater flexibility and control over the AI's behavior and tone.

I attempted to set up the "Custom" provider, but it still doesn't work.

I'm confident that the following curl command should run successfully.

I think it's a bug in LINQPad. I traced the request payload from LINQPad and the response content from Azure OpenAI Service. I realized that the first response in Stream mode will produce an "EMPTY" choices and LINQPad think that's an error.

Here is the very first response chunk of the response from AOAI.

{
    "choices": [],
    "created": 0,
    "id": "",
    "model": "",
    "object": "",
    "prompt_filter_results": [
        {
            "prompt_index": 0,
            "content_filter_results": {
                "hate": {
                    "filtered": false,
                    "severity": "safe"
                },
                "jailbreak": {
                    "filtered": false,
                    "detected": false
                },
                "self_harm": {
                    "filtered": false,
                    "severity": "safe"
                },
                "sexual": {
                    "filtered": false,
                    "severity": "safe"
                },
                "violence": {
                    "filtered": false,
                    "severity": "safe"
                }
            }
        }
    ]
}

There is a related issue here: https://github.com/Azure/azure-sdk-for-go/issues/21086

Also here: https://github.com/streamlit/streamlit/issues/8084

I think the best way to solve this issue is by ignoring empty choices chunk.

I'm sure my Azure OpenAI Service is able to use, but I just can't use in LINQPad.

Host runtime version: 8.0.15
Default query runtime version: 9.0.4
Default query reference assembly version: 9.0.4
Roslyn Version: 4.13.0-3.25120.11
FSharp.Compiler.Service version: 43.8.101.0
NuGet client version: 6.7.1.1
Windows DevDrive Status: Unavailable
Results rendering engine: Edge Chromium Engine 135.0.3179.73 (WebView2)

Hi,
we are interessted in purchasing LinqPad. Would it be possible to deploy LinqPad to a customer server if we purchase the enterprise edition?

A server would in general use the command-line version of LINQPad (lprun) or other automated scenarios which don't require licensed features. The Premium edition adds only UI features such as autocompletion.

Good point - there should be context menus items for "Copy database name" and "Copy connection string". I'll add this to the next build.

Regarding question 1, I had a look at the internals of LinqPad to see what is going on and I think this could be fixed with a slight performance hit.

At the moment, Linqpad builds a DataTable from the list of ExpandObjects without checking the type of each column and hence each DataColumn is a string.

https://share.linqpad.net/s5d5mn9v.linq is a benchmark script with an amended copy of this routine that sets the column type iff every non-null value for that column has the same type.

The benchmark runs against 10000 records (using the demo database supplied with Linqpad but any database could be used).

The results are as follows.

I think this performance hit is worth the benefit for the following reasons.

• It is only a 2% slowdown.
• It is only a couple of milliseconds when processing 10,000 records
• I would not be using ExpandObjects if I was that worried about performance.

It is relying on the fact that at the point where this code is used by LinqPad the source has been converted to a list, so even if the original source was IQueryable we are not going to be hitting the database several times. Hence my benchmark is using a list.

It is also relying on the fact that DataTable handles nulls, hence we do not need to worry about whether the columnType should be nullable or not.

Are there any downsides that I have not considered?

What do you think?