Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/zyga/ubuntu-update-alternatives-data

Database describing how update-alternatives is used by all packages in Ubuntu
https://github.com/zyga/ubuntu-update-alternatives-data

Last synced: 2 days ago
JSON representation

Database describing how update-alternatives is used by all packages in Ubuntu

Awesome Lists containing this project

README 

        
Ubuntu update-alternatives Data

===============================



This repository contains a automatically created and manually verified database

of how each package in Ubuntu (percise, quantal, raring) is using the

update-alternatives command.



The intent of this database is to help build reliable data for

command-not-found which suggests which packages contain a given command

executable



The long term solution is to move this data right into each package. Preferably

into the control file itself.



How much of the archive is covered?

===================================



This database covers about 11% of the Ubuntu archive.



How to generate this data?

==========================



This data is generated manually, by confirming automatic suggestions and fixing

things that were not handled automatically with manual edits.



To generate the data you need a full copy of the ubuntu archive (easily

obtainable with debmirror) and the check-update-alternatives script from the

command-not-found repository

(https://github.com/zyga/command-not-found/tree/check-update-alternatives).



Run the command as follows:



    ./check-update-alternatives /path/to/archive update-alternatives-db.ini



And answer questions as they are asked. Note that it is safe to interrupt the

program while a question is being asked. All previous answers are retained

and written back to the database.



Hot to interpret the data?

==========================



The database is saved as a windows-style ini file. Sections represent distinct

records. The name of each section is composed of the name and version of a

package. Since Debian package versions cannot be reused it is safe to assume

that each record uniquely identifies a particular binary package.



The most important data stored in each record is the ``alternatives`` key. If

present, it contains a space-separated list of pairs ``link=path`` that

enumerate all usages of update-alternatives of a given package.



The presence of the ``manual-check == yes`` key-value pair indicates that the

package needs to be inspected manually. This key should be removed after

inspection.



The presence of the ``using-hint = yes`` key-value pair indicates that this

record was derived from a previously validated guess. Technically the guess is

based on hints, not on the ``postinst`` script so this _might_ be incorrect but

usually is not. 



The presence and value of the ``hint-correct`` key encodes if the automatically

generated list of hints was correct or not, as determined by manual inspection.



The value of the ``hint-hash`` key encodes the hash of the canonical

representation of the list of hints. This is used as an implementation detail,

to accelerate discovery of hints that can be reused. See the source code for

details on how this is computed and used. 



The presence of the ``using-hint = yes`` key-value pair indicates that this

record reused an answer from another record and that it was not manually

verified.



The presence of the ``using-variables = yes`` key-value pair indicates that

hints contain shell variables.



The presence of the ``empty-hints = yes`` key-value pair indicates that despite

having ``update-alternatives`` call somewhere in the ``postinst`` script, the

list of hints was empty.