#!/usr/bin/env nix-shell
#!nix-shell -p sent -i sent
#
# vim: set spell:

Wiki-Member Berries
How I Learned to Stop Worrying and Love the PmWiki
\
oxzi @ hsmr, Siebter des Monats 2020-11

# Screenshot of hsmr.cc to illustrate the wiki.
@img/hsmr-screenshot.png

- Wiki exists for over seven years
- Thousands of edits
- Contains both memories and research questions

Originally planned: Something like Data Science
Therefore the data needs to be in a useful format

# Image macro to show ssh over ssh over ssh.
# 'cause ssh to türstatus to wega.
@img/ssh-head-over.png

PmWiki is the software and uses flat files
wiki.d directory full of PageFileFormat files

Each Wiki page is an own PageFileFormat file

Plain text file format consisting of key-value pairs
\
> version=pmwiki-2.1.0 urlencoded=1
> text=Markup text

PageFileFormat also contains its own history
\
> author:${TIME_NOW}=${AUTHOR}
> diff:${TIME_NOW}:${TIME_PREV}:=${DIFF}

# Sketch to show that patches are derived from the current state.
@img/page-file-format.png

Diff to previous versions is a classic diff (1)
\
0a1
> New first line, yay
3,4d3
< here goes my favourite line
< oh noes
23c23
< mostly used to fix mstks
--
> mostly used to fix mistakes

Overall it is quite easy t͝o͢ pars̵e

B̢u̸t ̢P͞m͟Wi͠k̸í ̵įs̶ a ̨s̕tan͝d̴álơn̴e͘ ̕a҉pplic͞atio̡n
W̛̕͟r͜͞it͠t̴e̡n͡ i̡̛n ̴̧̕a͞n̢̕c͞į͞e͏n̛҉t ̕҉͟PH͝҉̕P̴͠
W͏̷̢͡i̕͟͟͢͝t̵́́͝͞h͝͞ ̛́ń̸̸̶͘o҉̶̧͘͠ ̕͝҉t͏̨̛͟e҉͜s͡҉͞t̸̴̨s͞ ̶̶̛a͡͡ń̵̸͘͠d̸̷ ̢͞á̸̧ ̵͞͞c̨҉̴ù̧͡s̀͢͢͟ţ̷́̀o͏͜҉m̢ ͝͏̶̧͟p̸̶͟͢a̵̵̢͝t̶̛͘ç̴̷̨̛ḩ͝ ̵̨͘a͠͠͞ǹ͢d̕͜͡ ̨͟͝҉d̢͏i҉f̸̨҉f̸̛͞͡

🥺

# Screenshot of PmWiki's patch restore function.
@img/pmwiki-restore.png

P̕m̷Wi҉k̴i is a gift that keeps on giving…
- Double and empty diffs against its own revision
- Trimmed lines in diffs
- Invalid diffs
- Missing or truncated history

This is where I stop crying
And make a leap in time

# https://github.com/oxzi/pmwiki-pagefileformat-go
@img/pagefileformat-go.png

$ git log --oneline | wc -l
9162

$ git quick-stats -w
Git commits by weekday:
\
        day     sum
        Mon     1899    |████████████████
        Tue     1549    |█████████████
        Wed     1214    |██████████
        Thu     1366    |███████████
        Fri     1090    |█████████
        Sat     1155    |██████████
        Sun     889     |███████

$ git quick-stats -o
Git commits by hour:
\
        hour    sum
        00      356     |███
        01      307     |██
        02      169     |█
        03      107     |
        04      53      |
        05      74      |
        06      67      |
        07      121     |█
        08      173     |█
        09      228     |█
        10      248     |██
        11      338     |██
        12      330     |██
        13      384     |███
        14      414     |███
        15      452     |███
        16      536     |████
        17      503     |████
        18      685     |█████
        19      938     |████████
        20      928     |████████
        21      748     |██████
        22      644     |█████
        23      359     |███

$ git quick-stats -a | head
Git commits per author:
\
        3040  oleander      33.2%
        1604  binbash       17.5%
        1091  oxzi          11.9%
        624   towb          6.8%
        599   manu          6.5%
        421   2cv6          4.6%
        269   pim           2.9%
        175   pmwiki        1.9%
        167   bjoern        1.8%
        119   bo            1.3%

# Screenshot of gource showing the repository.
@img/gource-screenshot.png

No Data Science, just Meta Data Science.