| 1 |
Hear ye, formalistic entity definition fans! |
| 2 |
|
| 3 |
Aforementioned evil project to interpret the PMS docs and recast them |
| 4 |
unofficially as a go-to formal grammar has advanced to its next stage of |
| 5 |
world domination! (Complete no-show by the forecast army of worker-gnomes |
| 6 |
notwithstanding.) |
| 7 |
|
| 8 |
As of a few minutes ago, thanks to some quirky ruby, almost-correct* |
| 9 |
regular expressions can now be generated for each of the various |
| 10 |
identifiers, which could be used in any tools. |
| 11 |
|
| 12 |
From a transparency/maintenance perspective it may be preferable to |
| 13 |
generate these for any package-related tools since it means everyone is |
| 14 |
sourcing their grammar - without error prone manual processes - from the |
| 15 |
same single point of truth, removing an entire class of bugs. |
| 16 |
|
| 17 |
Code @ https://github.com/globalcitizen/gentoo-pms-abnf |
| 18 |
Related bug @ https://bugs.gentoo.org/show_bug.cgi?id=456892 (full of |
| 19 |
unbelievers) |
| 20 |
|
| 21 |
- Walter |
| 22 |
|
| 23 |
* 'Almost correct' because entities like "a" are not officially allowed in |
| 24 |
ABNF, but desirable in spec for readability and ease of editing. An |
| 25 |
intermediate script is thus needed prior to regex generation to simply |
| 26 |
convert strings in to escaped entities ('%xxx' IIRC) to get properly solid |
| 27 |
output. Easy one but I'm otherwise disposed today. |
Archives