POSTGRESQL PROJECTITS ORGANIZATION, ITS CONTRIBUTORS

Written by / Guillaume Lelarge @g_lelarge

AUTHORAuthor: Guillaume Lelarge

major contributorex - pgAdmin devfrench translator of the PostgreSQL manualtreasurer of PostgreSQL Europe

Company: Date: June 2016

Dalibo

AGENDAOrganization throughout PostgreSQL's history

Issues, and solutionsTools usedDevelopment

minor, and releasealpha, beta, release candidate, GA

Different kind of contributorsOrgs related to PostgreSQL

intl NPOs, National NPos, and meetups

PART 1EVOLUTION

POSTGRESQLFree Relational DataBase System (MIT/BSD)Very respectful of standards and dataGreat portabilityAwesome community

A BIT OF HISTORY1970: Ingres development1985: Postgres development1995: Adding SQL support1996: code is free

project renamed to PostgreSQLcreation of PGDG

DEV TOOLS IN 1996PostgreSQL Licence (MIT)CVS repositoryMailing lists

DEV CYCLE IN 96Patches sent on the mailing list... and commited when a commiter had the time to checkthe patchFew commiters

CHECKING A PATCHData consistencySQL standardPerformancesInteresting feature for many peopleCode simplicityDocumentation

THE COMMUNITY GROWSReleases are comingOne new major release every 10/12 monthsThe community grows

especially with the 8.0 and 9.0 releasesThe developper community grows too

PROBLEMS START COMING INMore developpers with less knowledge of the whole codeMore supported platforms

BUILDFARM TOOLWebapp to show results of compiling/installing/unittestingPerl client making easy to add a compilation nodeWritten and maintained by Andrew Dunstan

NEW PROBLEMS ARRIVEMore and more patchsBut still few committersHence a patchs waiting list growing:

some patches are checked very late in the dev cycledevelopers starting to be angryfinal releases being pushed backusers and commiters angry too

FIRST SOLUTION: COMMITFESTAt the beginning of the 8.3 dev cycleFor each new patch, add an entry in the wikiPatch reviewers working every too months to empty thepatch queuePatch reviewing is quickerBut many reviewers are afraid of big patches

HAVING A BETTER COMMITFEST8.4 development

add a commitfest handlercreate a commitfest webapp

9.0 developmentalpha release packaging

SECOND SOLUTION: MORECOMMITERS

4 new commiters during 9.0 dev cycle2 more during 9.3 dev cycle1 more during 9.6 dev cycleA total of 22 commiters

STILL MORE ISSUESMore unit testing in the build farmBut nothing performance-wise

NEW SOLUTION: PERFORMANCEFARM

Hard to get it rightStill a WIP

PART 2TOOLS

COMPILATION TOOLSMostly C Language

ISO89Standard Linux ToolchainWindows Toolchain

before 9.2, Visual C++ 20089.2, Visual C++ 20129.3 and 9.4, Visual C++ 2015since 9.5+, Visual C++ 2015

SOURCE REPOSITORY TOOLCVS from the beginning and till 2010CVS-Git for 2 yearsMoving to Git mi-september 2010

BUG TRACKER TOOLNo bugtracker for PostgreSQLpgsql-bugs for bugspgsql-* for feature requestsBut... we see some work on this during the last months

TALK TOOLAll discussions happen on mailing lists

mostlypgsql-hackers for developmentand many pgsql-* for the rest

COMMITFEST TOOLcommitfest.postgresql.org webappWritten by Robert HaasPHP language, PostgreSQL database (what else?)

COMMITFEST TOOL, V2commitfest.postgresql.org webappWritten by Magnus HaganderPython language, though still a PostgreSQL database

BUILDFARMMany servers (40)Many OSesMany compilatorsMany distributionsDifferent compilation options

PART 3THE VARIOUS STEPS OF AN UPGRADE

TWO KINDS OF RELEASESMinor: bug and security fixesMajor: new features

BUG FIXFirst check it actually is a bugFix the bugged releaseCheck on every other maintained releases

and the next oneAnd fix them if appropriate

NEW MINOR RELEASEHappens on all maintained stable releasesEvery 2/3 months

unless major bug (data corruption and alike) andsecurity issue

NEW MAJOR RELEASEEvery 12/15 monthsRight a�er, creation of a new branchDevelopment of the next major release starts

except for 9.1 and 9.2 where the development of thenext release started before the release of the currentone

PATCH LIFE CYCLEDiscussion

find the actual issueto search the right fix

Write the codeSubmit the patchReview the patchFix the patch if necessaryCommit the patch

ALPHA RELEASESA�er each commitfestAllow users to test new features without needing to grabthe git releaseAllow devs to know if a a bug in a specific alpha releasehas already been fixed... or notNot done anymore

FEATURE FREEZEMight be very longCheck latest patchesDebug, debug, debugTranslation workAdvocacy work

BETA RELEASESNo new featuresAdd available translationsSource packagingAnd some binary packagingCommon effort from users and devs

feature testscompilation testsperformance tests

RELEASE CANDIDATEOnly when there's no bugs le�Last not much

J-5: PACKAGESTag in the repositoryBuild source packageSend informations to the packagersPrepare binary packages

RELEASE DATEInternational press kitAnnounce on the website and on the mailing lists

PART 4CONTRIBUTORS

CORE TEAMFrom le� to right, and up to down...

Tom LaneBruce MomjianMagnus HaganderDave PageJosh BerkusPeter Eisentraut

CORE TEAM DUTIESChoosing the right date for a new releaseContact for security issuesDecide when the community doesn't reach a consensus

CODE CONTRIBUTORSDevReviewerCommiter

OTHER CONTRIBUTORSAll are very important

admin teamadvocacy teamspeakerstranslators

CONTRIBUTOR ON A FREE PROJECT:A HOBBY?

Some do it on their free timeBut most of them are paid for it

either by their companyor for a specific code writing

PART 5INTERNATIONAL COMMUNITY

MEETUPSIn every countriesItalyFrance (Paris, Lyon, Nantes, Lille)United KingdomSweden

NATIONAL ORGSJapanFranceItaliaRussia

MULTI-NATIONAL ORGSPostgreSQL EuropePostgreSQL US

SOLUTIONS LINUX 2007, JAN 2007

FOSDEM 2007

PGDAY.IT 2007

PGDAY.IT 2007

PGDAY.IT 2007

PGDAY.IT 2007

PGDAY.IT 2007

FOSDEM 2008

PGDAY.EU 2010

PGDAY.EU 2010

PGDAY.EU 2010

FOSDEM 2011

FOSDEM 2011

PGCONF.EU 2016

LIST OF NEXT EVENTS2016-07-05, PGDay UK 2016, London, United Kingdom2016-07-06, PGDay'16 Russia, St. Petersburg, Russia2016-09-13 – 2016-09-16, PostgresOpen 2016, Dallas,Texas, United States2016-11-01 – 2016-11-04, PGConf.EU 2016, Tallinn, Estonia

CONCLUSIONPostgreSQL is a project very much aliveTrying to find solutions to organization issuesTrying to incorporate every kind of contributors