Fabien Benetou's PIM | CognitiveEnvironments / Programming

Principle

Explicitly facilitate being in the most efficient environment to produce and continuously improve code.

Specific environments (to cycle through)

Clarifying the problem to solve

where is the value?
?

Architecture design

what is organization decision will have the biggest impact of the code maintanability and extensibility?
?

Implementing

do I understand the good practice of the language? environment? paradigm?
- do I apply them?
periodically compare own produced code to
- beginners, intermediate, veterans
?

Debugging

construct the most likely hypothesis
check available log to confirm it
1. if it is correct, modify the code to fix it
2. if it is not, modify the hypothesis and cycle again

Refactoring

does the code smell?
does code coverage shows that most of the code is rarely used thus should not be spent so much time on?
reconsider ArchitectureDesign, now that the code has been used what has changed and consequently how should the code reflect that?
?

CodeRefactoring displays the RSS to periodically propose refactoring to keep the code clean, yet tools could help for code smell, code coverage and usage.

Testing

how can I make sure that the code does what it should do?
- and only that?
?

Documenting

who will be the persons most interacting with the code?
- by usage? by maintenance? by integration? by extending it?
?

Improving process

Clarify the steps required for an efficient workflow and what can support them.

Git#WikiIntegration
Programming for general links
Testbed:. to facilitate tests
wiki integration
- generated Recipes (hook on remote post-commit)
- search Keywords#code
  - which might be interesting to compare with Programming#CodeReUse
- intermap Repository: to link to pieces of code even per lines or as searches
- CodeRefactoring via Wiki.MemoryRecipe and the repository feed
- explore glip PHP library since PmWiki is in PHP (note also the php-src git repository
  - discovered via http://redotheoffice.com/?p=130 , warning though, AFAIK it works only on bare repositories
hooks
- facilitating following a proposed paradigm by being integrated as the workflow
  - e.g. ΦFP (and the recently read WithoutNotesJuly11#GraphsBrainsAndGremlin which seems pretty coherent with it)
  - consider also integrating with different checkers for each layer of the stack e.g. PHP interpreter errors and warnings, Javascript engine, HTML validators, w3c linkcheck, ...
- quality checking
  - local pre-commit with software analyzers e.g. [[Tools/JavaScript files with jslint, frama-c or coccinelle, Simian (Similiarity Analyzer), Stylecop, Fxcop, ndepend
    - see also Wikipedia:Software analysis pattern in general and the Software Analytics Group at Microsoft Research Asia (overall dedicated research labs a la Sourcerer listed before, should be organized)
  - it might be better to do so earlier in the process, e.g. in the IDE like Vi
  - on the last commit that is adding lines check against code repositories e.g. Programming#CodeReUse (or directly tools listed there) using language vocabulary (rather than words in texts) and a technique e.g. LDA to suggest similar code to study, ideally labeled as of good quality snippets
    - even though it can be done via a hook, this might be problematic in the workflow, consequently this could be updating a document to review later on during a moment more adapted for learning
- automated deployment
  - local post-commit and remote post-receive hooks which
    - rsync the right files to the right locations
      - either manually by selecting locations in OwnWikisNetwork or by scanning which in the list has which plugins installed
      - notebooksof.ourp.im farm (currently deactivated)
    - verify that the configuration calling file is ok else update them
    - package a content-free self-explanatory structured and configured version
      - eventually with a small httpd and an installer
      - a nightly package might be more adapted though
    - log the result, eventually create an RSS for problems
    - maintain the other "distribution channels"
      - e.g. PmWiki official list of recipes with for each the page and eventually its attachment
        thus probably requiring a curl post request
    - generate screenshots
      - (if not working on a library) change in the interface in any way have to generate screenshots and associated thumbnails as people do not judge the result by internal changes or how the code repository is going but by how it looks
  - cf TheArtOfLeanSoftwareDevelopment#Chapter2

Methods and techniques

Notes on content

To do

check Wikipedia:Philippe Kahn#Philippe.27s_3_Rules_of_Software_Craftsmanship
read Sylvain Ross' Software production heuristics
consider Vim (or other tools) integration to be part of the workflow
periodically step back
1. e.g. read http://www.antoarts.com/10-ways-to-improve-your-programming-skills/
2. reorganize resources on this wiki, in particular Programming
add LeechBlock sites list URL

Programming {CognitiveEnvironments}