LO's build system builds all components, doesn't providing options to exclude one (or some) of them, not needed for user. That requests time to build and disk space (for now usually not critical, but may). For example — Gentoo GNU/Linux's default conf builds LO Base in unusable state (no back-ends available). Because of build system don't provide options to exclude it. Although some binary distros provides separate packages for LO Writer, LO Calc /etc… Downstream report: https://bugs.gentoo.org/show_bug.cgi?id=549548
This was discussed in the past several times. And always rejected: 1. We want less build complexity, not more. 2. It does not work that way anyway--various unit tests need the application libraries. Base might be an exception, but it not easily possible to build without Writer or Calc. Btw, all database-related functionality can be disabled by --disable-database-connectivity (unless it has bit-rotted).
(In reply to David Tardon from comment #1) > This was discussed in the past several times. Discussed — well. But what about proper documentating the results. Something like a link to FAQ's answer… > 1. We want less build complexity, not more. Why not want correctness and completeness first? > 2. It does not work that way anyway--various unit tests need the application > libraries. Base might be an exception, but it not easily possible to build > without Writer or Calc. Base NOT «might», but IS the clearest exception. Keeping in mind Linux default (using shared libraries) you'll get lot of wonders trying to make operable (in right way (!)) collaborate editing feature. > Btw, all database-related functionality can be disabled by > --disable-database-connectivity (unless it has bit-rotted). All including interface (i.e. Base)? For now I'm interested more in advanced build, than in disabling Base. But due to orienting on less complexity the only available database engine is built-in. That provides complexity while operaing with small databases, just few tables with maximum thousands of rows.