"Improve" now edits existing code base #721

UmerHA · 2023-09-19T10:15:23Z

Solves #650.

Improve mode (ie -i flag) now produces code changes and applies those changes to the existing codebase.

Changes:

improve prompt now produces code changes in the git conflict marker format
the LLM output is parsed into code changes (called "Edits")
Edits are applied to existing codebase (either create new file or update existing one)

Note: Unlike Aider (https://github.com/paul-gauthier/aider/blob/main/aider/coders/editblock_coder.py) which uses a lot of code to make sure edits are correctly parsed and applied, I have kept the code very simple. Still, I have found it to work well with an adjusted prompt. Still, I would encourage everyone to test this code.

…gineer into improvement-prompt

lukaspetersson · 2023-09-19T10:46:06Z

Did you also run it with 3.5-turbo? Some logic is dependant on LLM output and 3.5 might be worse at following this.

I will try it myself and let you know if I get issues.

UmerHA · 2023-09-19T12:02:53Z

Did you also run it with 3.5-turbo? Some logic is dependant on LLM output and 3.5 might be worse at following this.

I will try it myself and let you know if I get issues.

No, I didn't. But I also wouldn't use gpt3.5 for code generation.

I think we should assume people use gpt4 or similarly capable models.

Happy to hear @AntonOsika's thoughts

pbharrin · 2023-09-19T14:27:56Z

Wow, excellent work. super excited to see this in action.

…gineer into improvement-prompt

fabhed · 2023-09-20T17:16:24Z

Nice work @UmerHA !

Looking at the code, I think it would make sense to re-use the logic in parse_chat instead of the adding similar logic in parse_all_edits function.

This part of parse_chat would have to be refactored or even removed, as I don't think we should continuously replace the readme.

    # Get all the text before the first ``` block
    readme = chat.split("```")[0]
    files.append(("README.md", readme))

ATheorell · 2023-09-21T09:34:20Z

Running b73f793 for improving someone else's program, I get the error:
KeyError: "File 'src/server/util/API.js' could not be found in '/home/axel/Software/Opensource-Contribution-Leaderboard/workspace'" when it is trying to apply the diff. Is the reason that, so far, it only supports improving gpt-engineer generated softwares?

Happy to provide the lengthy stack trace if requested.

UmerHA · 2023-09-21T09:47:51Z

Running b73f793 for improving someone else's program, I get the error:
KeyError: "File 'src/server/util/API.js' could not be found in '/home/axel/Software/Opensource-Contribution-Leaderboard/workspace'" when it is trying to apply the diff. Is the reason that, so far, it only supports improving gpt-engineer generated softwares?

Without thinking too in-depth about it, I think it should work for other non-gtpe-generated code as well.

Happy to provide the lengthy stack trace if requested.

Yes please! Would be helpful!

ATheorell · 2023-09-21T14:41:56Z

(venv) axel@axel-ThinkPad-T470s:~/Software/Opensource-Contribution-Leaderboard$ gpt-engineer -i .
How do you want to select the files?

Use File explorer.
Use Command-Line.
Use previous file list (available at /home/axel/Software/Opensource-Contribution-Leaderboard/.gpteng/file_list.txt)

Select option and press Enter (default=3): 2
Opensource-Contribution-Leaderboard/
├── admin/
│ ├── build/
0. │ │ └── webpack.config.js
│ ├── dist/

│ │ ├── app.5f898dd63d28f1d54c0b.js
│ │ ├── assets/
│ │ │ ├── images/
│ │ │ │ └── loginBg.jpg
│ │ │ └── layer/
│ │ │ ├── layer.js
│ │ │ ├── mobile/
│ │ │ │ ├── layer.js
│ │ │ │ └── need/
│ │ │ │ └── layer.css
│ │ │ └── theme/
│ │ │ └── default/
│ │ │ ├── icon-ext.png
│ │ │ ├── icon.png
│ │ │ ├── layer.css
│ │ │ ├── loading-0.gif
│ │ │ ├── loading-1.gif
│ │ │ └── loading-2.gif
│ │ ├── favicon.ico
│ │ ├── index.html
│ │ └── vendor.248856e3852433d0f86e.js
│ ├── docs/
│ │ └── how-to-make-your-own-scaffold.md
│ ├── favicon.ico
│ ├── index.html
│ ├── LICENSE
│ ├── node_modules/
│ ├── package-lock.json
│ ├── package.json
│ ├── README.md
│ ├── src/
│ │ ├── assets/
│ │ │ ├── images/
│ │ │ │ └── loginBg.jpg
│ │ │ └── layer/
│ │ │ ├── layer.js
│ │ │ ├── mobile/
│ │ │ │ ├── layer.js
│ │ │ │ └── need/
│ │ │ │ └── layer.css
│ │ │ └── theme/
│ │ │ └── default/
│ │ │ ├── icon-ext.png
│ │ │ ├── icon.png
│ │ │ ├── layer.css
│ │ │ ├── loading-0.gif
│ │ │ ├── loading-1.gif
│ │ │ └── loading-2.gif
│ │ ├── index.js
│ │ └── style/
│ │ ├── noty.css
│ │ └── style.css
│ └── yarn.lock
├── archive/
├── build/
│ └── webpack.config.js
├── docker-compose.yml
├── Dockerfile
├── docs/
│ └── images/
│ ├── demo.png
│ └── logo.png
├── favicon.ico
├── index.html
├── LICENSE
├── memory/
│ └── logs/
├── node_modules/
├── package-lock.json
├── package.json
├── prompt
├── README.md
├── REST-API.md
├── src/
│ ├── assets/
│ │ ├── data/
│ │ │ ├── data.json
│ │ │ └── log.json
│ │ └── images/
│ │ └── rocket-chat.svg
│ ├── index.js
│ ├── server/
│ │ ├── admindata.json
│ │ ├── app.js
│ │ ├── config-example.json
│ │ ├── config.json
│ │ ├── package.json
│ │ ├── refresh.js
│ │ ├── tests/
│ │ │ ├── fetch.js
│ │ │ └── index.js
│ │ └── util/
│ │ ├── API.js
│ │ └── Util.js
│ └── style/
│ ├── bootstrap.css
│ └── style.css
├── tests/
│ ├── access.sh
│ └── api-call-tests.sh
└── workspace/
└── src/
└── server/
└── util/
Select files by entering the numbers separated by commas/spaces or
specify range with a dash.
Example: 1,2,3-5,7,9,13-15,18,20 (enter 'all' to select everything)

Select files:54,58,61

The following files will be used in the improvement process:
file_list.txt:
/home/axel/Software/Opensource-Contribution-Leaderboard/src/server/app.js
/home/axel/Software/Opensource-Contribution-Leaderboard/src/server/refresh.js
/home/axel/Software/Opensource-Contribution-Leaderboard/src/server/util/API.js

The inserted prompt is the following:
'make the program get the list of contributors directly from github, instead of reading from the local config'

You can change these files in your project before proceeding.

Press enter to proceed with modifications.

PLANNING:
The current implementation reads the list of contributors from a local configuration file. To fetch the list of contributors directly from GitHub, we need to use the GitHub API.

The GitHub API provides an endpoint to fetch contributors for a specific repository. The endpoint is https://api.github.com/repos/{owner}/{repo}/contributors. We need to make a GET request to this endpoint to fetch the list of contributors.

We will create a new function getContributors in the API.js file. This function will take organization and repository as parameters and will return a list of contributors for the given repository.

Then, we will modify the getAllContributorsInfo function in the refresh.js file to use this new getContributors function instead of reading from the local configuration file.

OUTPUT:

src/server/util/API.js
<<<<<<< HEAD
module.exports = {
    getRepositories,
    getContributorAvatar,
    getOpenPRsNumber,
    getMergedPRsNumber,
    getIssuesNumber,
    getContributorInfo,
    checkRateLimit,
    getStats,
    getRanks,
}
=======
async function getContributors(organization, repository) {
    const res = await get(
        APIHOST + `/repos/${organization}/${repository}/contributors`
    )
    if (res !== undefined) {
        return res.data.map((element) => {
            return element['login']
        })
    } else {
        return []
    }
}

module.exports = {
    getRepositories,
    getContributorAvatar,
    getOpenPRsNumber,
    getMergedPRsNumber,
    getIssuesNumber,
    getContributorInfo,
    checkRateLimit,
    getStats,
    getRanks,
    getContributors,
}
>>>>>>> updated

src/server/refresh.js
<<<<<<< HEAD
let Config = jsonfile.readFileSync(configPath)
let organization = Config.organization
let contributors = Config.contributors
let includedRepositories = Config.includedRepositories
=======
let Config = jsonfile.readFileSync(configPath)
let organization = Config.organization
let includedRepositories = Config.includedRepositories
let contributors = []
for (let repository of includedRepositories) {
    let repoContributors = await API.getContributors(organization, repository)
    contributors = [...contributors, ...repoContributors]
}
contributors = [...new Set(contributors)] // remove duplicates
>>>>>>> updated
```Traceback (most recent call last):

  File "/home/axel/Software/gpt-engineer/venv/bin/gpt-engineer", line 8, in <module>
    sys.exit(app())

  File "/home/axel/Software/gpt-engineer/venv/lib/python3.10/site-packages/gpt_engineer/main.py", line 98, in main
    messages = step(ai, dbs)

  File "/home/axel/Software/gpt-engineer/venv/lib/python3.10/site-packages/gpt_engineer/steps.py", line 366, in improve_existing_code
    overwrite_files_with_edits(messages[-1].content.strip(), dbs)

  File "/home/axel/Software/gpt-engineer/venv/lib/python3.10/site-packages/gpt_engineer/chat_to_files.py", line 150, in overwrite_files_with_edits
    apply_edits(edits, dbs.workspace)

  File "/home/axel/Software/gpt-engineer/venv/lib/python3.10/site-packages/gpt_engineer/chat_to_files.py", line 207, in apply_edits
    workspace[filename] = workspace[filename].replace(

  File "/home/axel/Software/gpt-engineer/venv/lib/python3.10/site-packages/gpt_engineer/db.py", line 64, in __getitem__
    raise KeyError(f"File '{key}' could not be found in '{self.path}'")

KeyError: "File 'src/server/util/API.js' could not be found in '/home/axel/Software/Opensource-Contribution-Leaderboard/workspace'"

ATheorell · 2023-09-22T09:06:34Z

I guess my error above is the one that is addressed in #728 @RareMojo ?

joenas · 2023-09-22T13:53:47Z

@UmerHA Is the idea that it will always improve existing files, not add new ones?

I'm a bit confused with how gpt-engineer is supposed to work in cases like mine. I have a generated project that I want to add more features to, not necessarily changing existing files only but also adding new ones.
If I do gpt-engineer [mydir] with existing files and ask it to add a new feature it will happily wipe my entire workspace.

As for your change I think it's awesome but that it also needs to be able to add new files. In my case I wanted to add more database models but GPT will output everything in the

<<<<<<< HEAD
=======
>>>>>>> updated

format, even new files.

Also if GPT happens to add some other output to the reply, like bash commands, we get a ValueError

    OUTPUT:
    ```bash
    # In your terminal
    npm install package
    ```

RareMojo · 2023-09-22T14:25:30Z

I guess my error above is the one that is addressed in #728 @RareMojo ?

Can't dig into it for a while, but I'm inclined to say no. It looks like the same fix I suggested is implemented in apply_edits already here.

ATheorell · 2023-09-22T14:48:30Z

@RareMojo , I think you are right, these are different things.

UmerHA · 2023-09-22T15:57:40Z

(venv) axel@axel-ThinkPad-T470s:~/Software/Opensource-Contribution-Leaderboard$ gpt-engineer -i . How do you want to select the files?
...
File "/home/axel/Software/gpt-engineer/venv/lib/python3.10/site-packages/gpt_engineer/db.py", line 64, in getitem
raise KeyError(f"File '{key}' could not be found in '{self.path}'")

KeyError: "File 'src/server/util/API.js' could not be found in '/home/axel/Software/Opensource-Contribution-Leaderboard/workspace'"

I can't seem to reproduce your error. Would it be possible for you to send me the entire project files, e.g. via Discord?
Of course only if the code is not sensitive in any way.

UmerHA · 2023-09-22T16:02:38Z

@UmerHA Is the idea that it will always improve existing files, not add new ones?

No, it can also add new files. In this case the before part (between HEAD and =======) will be empty.

I'm a bit confused with how gpt-engineer is supposed to work in cases like mine. I have a generated project that I want to add more features to, not necessarily changing existing files only but also adding new ones. If I do gpt-engineer [mydir] with existing files and ask it to add a new feature it will happily wipe my entire workspace.

When you do gpt-engineer -i [mydir] it should only update the files you named, or create new ones. Can you describe in more detail under which circumstances it wipes out existing files?

Also if GPT happens to add some other output to the reply, like bash commands, we get a ValueError
    OUTPUT:
    ```bash
    # In your terminal
    npm install package
    ```

Could you post the 'ask' you made to gpte? That helps me understand better if we need to add handling of such cases. :)

ATheorell · 2023-09-23T07:03:45Z

Here is the project that I'm trying to modify: https://github.com/ATheorell/Opensource-Contribution-Leaderboard/tree/improveCode

The prompt and the file selection are in the stack trace. I tried multiple times and always get the error @UmerHA

SabareeshGC · 2023-09-25T16:35:08Z

I am using this for java and but several times it is not actually updating the files and i dont see any error logs. Trying to narrow down but no luck

UmerHA · 2023-09-25T19:18:29Z

Here is the project that I'm trying to modify: https://github.com/ATheorell/Opensource-Contribution-Leaderboard/tree/improveCode

The prompt and the file selection are in the stack trace. I tried multiple times and always get the error @UmerHA

Okay, can reproduce now. The issue is that currently gpte expects the code to be in a subfolder name 'workplace'. I'll make the name of the workplace folder editable. In your case, you would then choose 'src'.

ATheorell · 2023-09-26T08:31:16Z

Here is the project that I'm trying to modify: https://github.com/ATheorell/Opensource-Contribution-Leaderboard/tree/improveCode
The prompt and the file selection are in the stack trace. I tried multiple times and always get the error @UmerHA

Okay, can reproduce now. The issue is that currently gpte expects the code to be in a subfolder name 'workplace'. I'll make the name of the workplace folder editable. In your case, you would then choose 'src'.

What is the downside of having it work with relative paths from the execution path, rather than an explicit code path? I'm thinking about the case that the improved code files may be scattered over the file tree (and also a bit about the general UX).

UmerHA · 2023-09-26T18:34:49Z

Here is the project that I'm trying to modify: https://github.com/ATheorell/Opensource-Contribution-Leaderboard/tree/improveCode
The prompt and the file selection are in the stack trace. I tried multiple times and always get the error @UmerHA

Okay, can reproduce now. The issue is that currently gpte expects the code to be in a subfolder name 'workplace'. I'll make the name of the workplace folder editable. In your case, you would then choose 'src'.

What is the downside of having it work with relative paths from the execution path, rather than an explicit code path? I'm thinking about the case that the improved code files may be scattered over the file tree (and also a bit about the general UX).

Do you mean project path (not execution path)? I mostly use gpte like this

$ cd path/to/gpte/venv
$ pipenv shell
$ gpt-engineer -i path/to/project

So, the execution path (path/to/gpte/venv) and project path (path/to/project) are different.

In principle, there is no downside to using relative locations to the project path. We would then have make sure gpte-internal files are not editable.

Iirc, there's a proposal to move all of those things into a .gpt-engineer folder, right? We could then just exclude that folder.

UmerHA · 2023-09-29T11:41:49Z

I think we need to decide how we want to structure the gpte directory. We have two options:

root dir is for gtpe, a subdir (eg "workspace") is for the code
root dir is for code, a subdir (eg ".gptengineer") is for gpte

We currently do 1, but when we want the ability to edit existing codebases, 2 would be easier. @AntonOsika @ATheorell @pbharrin thoughts?

When that's decided, then I can finish this PR.

UmerHA · 2023-10-01T09:34:52Z

I now went ahead and made a PR to use the project path as workplace and move everything else into .gpteng - see #749. That would need to be merged first, to fix & merge this PR.

@AntonOsika @ATheorell

AntonOsika

Awesome.

Looking forward to try.

The prompt is pretty long, so I'd also like to compare it with sweep.dev's approach:)

I added some comments, feel free to merge without addressing, but nice if we at least address after merge!

gpt_engineer/chat_to_files.py

gpt_engineer/main.py

gpt_engineer/preprompts/ht.txt

tests/test_chat_to_files.py

AntonOsika · 2023-10-02T07:11:44Z

Good job, check my comments and also tests + pre-commit is failing:

UmerHA · 2023-10-04T09:12:28Z

The prompt is pretty long, so I'd also like to compare it with sweep.dev's approach:)

@AntonOsika Agree the prompt is long & we should look into making it shorter. I would, however, prefer to separate it from this PR. Let's get this shipped, and then optimize it later.

I added some comments, feel free to merge without addressing, but nice if we at least address after merge!

Addressed all!

tests + pre-commit is failing

works now

Also, reminder we need to merge #749 first :)

pbharrin

Looks great, thanks for updating the tests.

UmerHA · 2023-10-12T11:13:57Z

resolved merge conflicts, all tests pass, Peter has approved; let's merge.

rrmistry · 2023-11-02T09:56:06Z

Has anyone tested scenario where the model suggests multiple edits within the same file using the -i|--improve flag?

Found that only the first model-suggested edit is made in the file, the following edits are not made in the file as expected.

Click to expand console log

Tested using Docker on main branch as of Nov 2, 2023 (commit: 1408652)

Command line:

export OPENAI_API_KEY="sk-2kMT........................................AXRK"
export DOCKER_USER="$(id -u):$(id -g)"
export OPENAI_MODEL="gpt-4"

docker run -it --rm -e OPENAI_API_KEY="$OPENAI_API_KEY" -v "/.../my-blazor-pwa:/project:cached" --user "$DOCKER_USER" gpt-engineer -i $OPENAI_MODEL | tee ".gpteng/logs/output_`date +"%Y%m%d_%H%M%S"`.txt"

With output:

Running gpt-engineer in /project

File list detected at /project/.gpteng/file_list.txt. Edit or delete it if you want to select new files.

The following files will be used in the improvement process:
file_list.txt:
�[32m/project/Server/Data/MyCloudWebAppDbContext.cs
�[0m

The inserted prompt is the following:
�[32mCan you convert the existing code that depends on Neo4j over to JanusGraph in C#? The client library is Gremlin.Net. The endpoint is available at the environment variable "MY_WEBAPP_JANUSGRAPH_ENDPOINT" with has the value "dev-my-cloud-janusgraph:8182" which should be dynamically parsed and split based on ":" character. Make sure that all the Graph Queries are updated to use the new endpoint.
Make sure that the queries are also translated to Gremlin equivalent. Both for querying as well as for mutations. The queries are already in-line in Neo4j client format and in comments. These need to be converted to Gremlin queries.
�[0m

You can change these files in your project before proceeding.

Press enter to proceed with modifications.

PLANNING:
The task involves converting the existing code that depends on Neo4j over to JanusGraph in C#. The client library is Gremlin.Net. The endpoint is available at the environment variable "MY_WEBAPP_JANUSGRAPH_ENDPOINT" with has the value "dev-my-cloud-janusgraph:8182" which should be dynamically parsed and split based on ":" character.

The steps involved in this task are:

Replace the Neo4j client initialization with the Gremlin.Net client initialization.

Replace all the Neo4j Cypher queries with equivalent Gremlin queries.

Update the method signatures and implementations to use the Gremlin.Net client.

OUTPUT:
Due to the complexity of the task and the length of the code, it's not feasible to provide a complete solution here. However, I will provide a few examples of how to convert the code.
Server/Data/MyCloudWebAppDbContext.cs
<<<<<<< HEAD
    private BoltGraphClient _neo4jClient { get; }
=======
    private GremlinClient _gremlinClient { get; }
>>>>>>> updated
Server/Data/MyCloudWebAppDbContext.cs
<<<<<<< HEAD
    _neo4jClient = new BoltGraphClient(
        uri: EnvironmentExtensions.GetEnvironmentVariableOrDefault(
            envVarName: "MY_WEBAPP_NEO4J_ENDPOINT",
            defaultValue: "bolt://my-cloud-neo4j:7687"
        ),
        username: neo4j_auth[0],
        password: neo4j_auth[1],
        encryptionLevel: Neo4j.Driver.EncryptionLevel.Encrypted
    );
=======
    var endpoint = EnvironmentExtensions.GetEnvironmentVariableOrDefault(
        envVarName: "MY_WEBAPP_JANUSGRAPH_ENDPOINT",
        defaultValue: "dev-my-cloud-janusgraph:8182"
    ).Split(":");
    _gremlinClient = new GremlinClient(new GremlinServer(endpoint[0], int.Parse(endpoint[1])));
>>>>>>> updated
Server/Data/MyCloudWebAppDbContext.cs
<<<<<<< HEAD
    var user = await _neo4jClient.Cypher
        .WithParam("id", userId)
        .Match("(u:MyCloudUser {id: $id})")
        .Where((MyCloudUser u) => u.Id == userId)
        .Return(u => u.As<MyCloudUser>())
        .Limit(1)
        .ResultsAsync;
=======
    var user = await _gremlinClient.SubmitAsync<MyCloudUser>("g.V().hasLabel('MyCloudUser').has('id', userId).limit(1)");
>>>>>>> updated
Please note that the above examples are just a starting point. The actual conversion might require more changes depending on the complexity of the queries and the specific requirements of the JanusGraph database.Total api cost: $ ...

I've confirmed that only the first edit was made in the target file.

UmerHA · 2023-11-16T13:45:33Z

@ATheorell pinging you so @rrmistry's comment doesn't get lost. I don't have time to fix this, but wanted to let the team know.

UmerHA added 10 commits September 18, 2023 10:38

check in 23-09-18

0e167fb

Merge branch 'AntonOsika:main' into improvement-prompt

273cb6c

Simplified parse+apply edits - still to test!

270dd5b

Merge branch 'improvement-prompt' of https://github.com/UmerHA/gpt-en…

85cd704

…gineer into improvement-prompt

Merge branch 'AntonOsika:main' into improvement-prompt

ba212e7

Changed improve prompt

bc3f938

Merge branch 'improvement-prompt' of https://github.com/UmerHA/gpt-en…

09087a3

…gineer into improvement-prompt

Fix: Saving new files

677af0a

Removed debug message

af6fd91

formatting

552eaa3

UmerHA added 3 commits September 20, 2023 00:12

Merge branch 'main' into improvement-prompt

89f19fa

Formatting

a50c772

Merge branch 'improvement-prompt' of https://github.com/UmerHA/gpt-en…

b73f793

…gineer into improvement-prompt

UmerHA mentioned this pull request Sep 22, 2023

Ask to run code after improvement #722

Closed

UmerHA mentioned this pull request Oct 1, 2023

Project path is now workspace #749

Merged

Merge branch 'AntonOsika:main' into improvement-prompt

7fa8d91

UmerHA requested a review from pbharrin as a code owner October 1, 2023 09:29

Integrated PR AntonOsika#749

0953bb0

AntonOsika reviewed Oct 2, 2023

View reviewed changes

UmerHA added 5 commits October 3, 2023 21:05

Merge branch 'gpteng-subfolder' into improvement-prompt

35864e0

Formatting

42350db

Incorporated review feedback

67b0e18

Merge branch 'main' into improvement-prompt

c6eb292

Formatting

ca10d46

pbharrin approved these changes Oct 10, 2023

View reviewed changes

UmerHA added 2 commits October 12, 2023 12:53

Merge branch 'main' into improvement-prompt

f76db3b

final run test

e438760

UmerHA merged commit c7f2d0d into AntonOsika:main Oct 12, 2023
4 checks passed

UmerHA deleted the improvement-prompt branch October 12, 2023 11:14

ATheorell mentioned this pull request Nov 25, 2023

Better edit syntax for the improve command #869

Closed

similato87 mentioned this pull request Feb 11, 2024

Git diff-style approach for improve function #1005

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

"Improve" now edits existing code base #721

"Improve" now edits existing code base #721

UmerHA commented Sep 19, 2023

lukaspetersson commented Sep 19, 2023

UmerHA commented Sep 19, 2023

pbharrin commented Sep 19, 2023

fabhed commented Sep 20, 2023

ATheorell commented Sep 21, 2023

UmerHA commented Sep 21, 2023

ATheorell commented Sep 21, 2023

ATheorell commented Sep 22, 2023

joenas commented Sep 22, 2023

RareMojo commented Sep 22, 2023 •

edited

Loading

ATheorell commented Sep 22, 2023

UmerHA commented Sep 22, 2023

UmerHA commented Sep 22, 2023 •

edited

Loading

ATheorell commented Sep 23, 2023

SabareeshGC commented Sep 25, 2023

UmerHA commented Sep 25, 2023

ATheorell commented Sep 26, 2023

UmerHA commented Sep 26, 2023 •

edited

Loading

UmerHA commented Sep 29, 2023

UmerHA commented Oct 1, 2023

AntonOsika left a comment

AntonOsika commented Oct 2, 2023 •

edited

Loading

UmerHA commented Oct 4, 2023

pbharrin left a comment

UmerHA commented Oct 12, 2023

rrmistry commented Nov 2, 2023 •

edited

Loading

File list detected at /project/.gpteng/file_list.txt. Edit or delete it if you want to select new files.

UmerHA commented Nov 16, 2023

"Improve" now edits existing code base #721

"Improve" now edits existing code base #721

Conversation

UmerHA commented Sep 19, 2023

lukaspetersson commented Sep 19, 2023

UmerHA commented Sep 19, 2023

pbharrin commented Sep 19, 2023

fabhed commented Sep 20, 2023

ATheorell commented Sep 21, 2023

UmerHA commented Sep 21, 2023

ATheorell commented Sep 21, 2023

Select files:54,58,61

The inserted prompt is the following: 'make the program get the list of contributors directly from github, instead of reading from the local config'

ATheorell commented Sep 22, 2023

joenas commented Sep 22, 2023

RareMojo commented Sep 22, 2023 • edited Loading

ATheorell commented Sep 22, 2023

UmerHA commented Sep 22, 2023

UmerHA commented Sep 22, 2023 • edited Loading

ATheorell commented Sep 23, 2023

SabareeshGC commented Sep 25, 2023

UmerHA commented Sep 25, 2023

ATheorell commented Sep 26, 2023

UmerHA commented Sep 26, 2023 • edited Loading

UmerHA commented Sep 29, 2023

UmerHA commented Oct 1, 2023

AntonOsika left a comment

Choose a reason for hiding this comment

AntonOsika commented Oct 2, 2023 • edited Loading

UmerHA commented Oct 4, 2023

pbharrin left a comment

Choose a reason for hiding this comment

UmerHA commented Oct 12, 2023

rrmistry commented Nov 2, 2023 • edited Loading

File list detected at /project/.gpteng/file_list.txt. Edit or delete it if you want to select new files.

UmerHA commented Nov 16, 2023

The inserted prompt is the following:
'make the program get the list of contributors directly from github, instead of reading from the local config'

RareMojo commented Sep 22, 2023 •

edited

Loading

UmerHA commented Sep 22, 2023 •

edited

Loading

UmerHA commented Sep 26, 2023 •

edited

Loading

AntonOsika commented Oct 2, 2023 •

edited

Loading

rrmistry commented Nov 2, 2023 •

edited

Loading