Taco Bell Programming (WIP)

• What are the core tenants of this style of programming, How do you define it ->
• How is it different from an imperative programming language ? (like Python)
• Can I overcome the limitations ? , if so how ? ->
• Figure out if i can code new projects faster with this style ->
• If so, how should I lay out code for different projects, in line with this style of programming
• When does this style work ? and when does it not ? (Usecases | Limitations)
• How to do error reporting ?
• How to log ? (syslog ?)
• How to do profilling / monitoring ?
• What language should i use to write tacoscripts , bash or haskell ? ->
• A standard for writing new tools to extend this community ->

What are the core tenants

• Write multiple reusable programs which are atomic in terms of its tasks, and can be slapped together to do something bigger
  • i.e parsing programs, should never do anything else, aka grep, sed, cut, awk, etc.
  • i.e data extraction from programs from raw pile should be seperate, aka find, curl, lynx, etc.
  • i.e action programs, should be seperate, aka notify-send, sendmail, alert, touch, mkdir, rm, etc.
• Should data be sent both ways ?
  • If we allow it, a lot of interaction between programs get abstracted away.
  • programs might lose modularity. i.e some programs might be incompatible bi-directionally with others
  • Only necessary in complex usecases.
• Limitations of not sending data both ways ?
  • An ingester digesting content cannot modify its behaviour based on data being parsed,
    by a program down the line
  • Hard to represent interactions, where user wants an updated status based on a particular interaction\ example:- A user submits an email form, and wants instant reply, shell scripting is prolly a bad idea for this
  • Further splitting of codebases.
    Code inter-related to code expressed in tacostyle, might need to be written in python3.
    This causes the programmer to refer to both places to figure out how a business logic works.
  • Security Nightmare Shell escape is a real threat

Can I overcome it ?

Tools that can aide me with this

• A tool to auto pull cheatsheet based on context, an interactive interface where if i type *.py, all python cheatsheets are shown if i type fastapi, fuzzy search is used to display while typing and then all specific cheatsheets are displayed (design this taco style)
• A tool to annotate items with data, systemfiles, urls, etc and store it and make it surfaceable when needed ( just use mkfifo )
• Use the custom job queue i built in Postgres as a mkfifo alternative
• A cli tool with gitblame to ask the correct person answer to some questions
• A suggestions tool, runs as a cron job, pulls whatever data it needs to and builds out the suggestion list
• A tool to copy output from to a central store, which can be extracted again to stdout, by other programs, using clicommand (storeto) do i need this ? (use mkfifo)
• DO NOT USE dnpipes this or any similar implementation, as multiple readers to a pipe is PAIN, just use kafka, and at max, stream it out to localnamed pipes using python3 scripts

Can I code faster with this style ?

Build a new page template

cat hugopage.md | ,grep-md  > mdpage.md
cat hugopage.md | ,grep-yaml | md2html mdpage.md | ,appendCSS tufte > finalpage.html

Send an email to all subscribers in an email list and update blacklist

cat subscribers.txt | sort | uniq | ,filter --csv --json --plain blacklist | ,toCSV subscribers.csv
cat emailcontent.md | toEML --assert-dir "assets/" --from "teito@teitoklien.com" --to-list subscribers.csv --outdir "shortcuts29th-email/"
mkfifo emailfeedback
awsfeedbackapp > emailfeedback 
cronit @5mins: cat emailfeedback | ,genBlackList --pre oldlist > newlist.txt && ,mv newlist blacklist 
# parsefeedback and update blacklist every 5 mins

The Standard

• All new cli tools must have ability to interface with storeto reserve directly to catch parameters
• Use && to concat independent actions
• Use tee and process substitution when necessary (refer here)
• Use namedpipes using mkfifo, whenever necessary to combine programs running and interfacing independently\ e.g: building queues/buffer using mkfifo
• All personally made/maintained cli tool name should start with a ,
• On passing no param and no stdin, Usage should be displayed
• All personally made tools, should have a man page displayed
• All commands should have a –dry option to not do anything but just output what it will do
• Use environment variables cautiously , and try to be explicity in cmdline itself instead
• Combine it with literate programming (refer here)
• Use makefile to further extend this style of shell scripting (refer here)

What language should i use

• This new pipe library here tutorial
• This literate programming lib here
• Mindfullness while coding, avoid repeating yourself

Other notes

• post by teddziuba here
• another post on makefiles by teddziuba here
• python and pipes unix series by lyceum here

Contact/etc