# Log 2023/08/2023-08-05
Going through old sketchbooks, writing down [[GO GO GOLEMS]] project ideas.
- [[IDEA - GO GO GOLEMS - 2023-08–05 - Prompt Context Manager]]
- [[IDEA - GO GO GOLEMS - 2023-08-05 - Reggie, regexp query and replacement program]]
- [[IDEA - GO GO GOLEMS - 2023-08-05 - flour, expose commands as services]]
- [[PROJ - GO GO GOLEMS - PARKA]]
- adding notes about exposing as RPC server, doing introspection and service the command itself, using the CLI to trigger the remote API in parka (using introspection first)
## Using tasks in [[Obsidian]]
For my [[My Obsidian workflow]], I want to tag incoming [[Readwise]] notes with two tasks: toFile and toProcess, so that I can make sure I process stuff I've read correctly. Using the [[Tasks Plugin (Obsidian)]], I can actually display all the not done tasks from the last week (no need to go further back to not overwhelm, it's already enough as is).
I used the following icons to have intermediate steps:
- https://publish.obsidian.md/slrvb-docs/ITS+Theme/Alternate+Checkboxes
## Generating [[GO GO GOLEMS]] commands
I used the following prompt to create the gtm [[go-go-labs]] tool:
- [[PROMPT - 2023-08-05 - Create glazed commands]]
This created the [[IDEA - GO GO GOLEMS - 2023-08-05 - Reggie, regexp query and replacement program|Reggie]] regex command:
```go
type RegexpMatchCommand struct {
description *cmds.CommandDescription
regexMap map[string]*regexp.Regexp
}
func NewRegexpMatchCommand(regexMap map[string]*regexp.Regexp) (*RegexpMatchCommand, error) {
glazedParameterLayer, err := settings.NewGlazedParameterLayers()
if err != nil {
return nil, errors.Wrap(err, "could not create Glazed parameter layer")
}
return &RegexpMatchCommand{
description: cmds.NewCommandDescription(
"regexpMatch",
cmds.WithShort("Regexp Match command"),
cmds.WithArguments(
parameters.NewParameterDefinition(
"inputFile",
parameters.ParameterTypeString,
parameters.WithHelp("Input file to apply regexes"),
parameters.WithRequired(true),
),
),
cmds.WithLayers(
glazedParameterLayer,
),
),
regexMap: regexMap,
}, nil
}
func (c *RegexpMatchCommand) Description() *cmds.CommandDescription {
return c.description
}
func (c *RegexpMatchCommand) Run(
ctx context.Context,
parsedLayers map[string]*layers.ParsedParameterLayer,
ps map[string]interface{},
gp middlewares.Processor,
) error {
inputFile := ps["inputFile"].(string)
file, err := os.Open(inputFile)
if err != nil {
return err
}
defer func(file *os.File) {
_ = file.Close()
}(file)
scanner := bufio.NewScanner(file)
for scanner.Scan() {
line := scanner.Text()
for name, regex := range c.regexMap {
matches := regex.FindStringSubmatch(line)
if matches != nil {
row := types.NewRow(
types.MRP("regex", name),
)
for i, match := range matches[1:] {
row.Set("group"+strconv.Itoa(i+1), match)
}
if err := gp.AddRow(ctx, row); err != nil {
return err
}
}
}
}
if err := scanner.Err(); err != nil {
return err
}
return nil
}
```
## IDEAS
I'm actually trying to:
- finish [[PROJ - GO GO GOLEMS - MASTOID]]
While doing this, I came across [[Omnivore]], which is a tool to manage snippets and the like, and apparently has the option to be self-hosted and can save full pages (which you can't export out of [[Readwise]]). This would be a good option to build the starting point of [[IDEA - A federated custom search engine]]. This has a section about self-hosting:
- https://blog.omnivore.app/p/deploying-a-minimal-self-hosted-omnivore
While doing so, I had the idea to add a tool to [[PROJ - GO GO GOLEMS - PARKA|Parka]] and [[IDEA - GO GO GOLEMS - 2023-08-05 - flour, expose commands as services]]
### Quickwrite ideas
- glazed [[JSON schema]] export and converter
- compile command to [[WASM]] and run in browser
- [[PROJ - GO GO GOLEMS - GLAZED]] server that can be used to run CLI commands remotely, but maybe that's just part of [[PROJ - GO GO GOLEMS - PARKA|Parka]] as a library feature.
- what does it mean to have plugins in [[GO GO GOLEMS]]
## [[PROJ - GO GO GOLEMS - MASTOID]]
- Added threaded view. Turns out I didn't have to iterate to get all the siblings, you get all the siblings moving down from the requested status
- Added the first release to fury.io, seems to have been pretty easy!
- More ideas:
- make it a general purpose data querying tool for mastodon
## Tasks for RWISE import
I want to have a better overview of the RWISE links I have already processed, because I read so much that I feel a lot of interesting articles get lost, not because I forgot them (nothing I can do about that), but because they aren't linked in the right places. I want to add tasks to each import so that I know the status of the different imports and possibly other types of notes as well.
I feel it is time to step up the organizational workflow. I actually wonder if I made ZK about that.
Indeed I have:
- [[ZK - 2023-07-30 - Making a workflow fun is the best way to make it stick]]
## Working on
While I'm at it, cleaning up ZK about [[🧠 Notetaking]] and creating:
- [[SK - 4a0 - Notetaking for Programmers]]
- [[SK - 4a0a - Building a Notetaking workflow]]
- [[SK - 2023-08-05 - My Own Thoughts]]
Sorting many ZK into their respective folders, in order to get the inbox down to a more reasonable size.
## Thoughts about [[SK - 2n6 - LLMs and Opensource]]
- [What (if any) machine learning data provenance practices does the Nivenly Foundation support? · nivenly/community · Discussion #2 · GitHub](https://github.com/nivenly/community/discussions/2#discussioncomment-6647878)
## Read
- [[RWISE - 2023-08-05 - ‘It’s Destroyed Me Completely’ Kenyan Moderators Decry Toll of Training of AI Models]]
- [[RWISE - 2023-08-05 - In Praise of Boring AI]]
- [[RWISE - 2023-08-05 - Consciousness, Artificial Intelligence, and Our Search for Meaning Oliver Sacks on ChatGPT, 30 Years Before ChatGPT]]
- [[RWISE - 2023-06-19 - We Need More of Richard Stallman, Not Less]]
- Filled out a bunch of AI stuff in [[Copyleft]] and [[Opensource Software]]
- [[RWISE - 2023-08-04 - A Tale of Two Snippets What Link Attribution in SGE Tells Us About Search]]
## Notes from phone
- Practice writing and publishing smaller articles
- Write one about pressing regenerate and how that exposes that creation is also curation