1. 07 Mar, 2018 1 commit
  2. 04 Feb, 2018 1 commit
    • Davor Ocelic's avatar
      Fix warning on make install · 12c8c47f
      Davor Ocelic authored
      Fixes a warning on 'make install' which is printed if the previous binary
      does not exist.
      
      Functionality the previous and current behavior are almost identical, with
      the benefit that the irrelevant error message is now silently ignored
      rather than too verbosely printed and potentially confusing the user.
      12c8c47f
  3. 09 Jan, 2018 2 commits
    • Davor Ocelic's avatar
      Minor change in Makefile (#496) · d38d3709
      Davor Ocelic authored
      d38d3709
    • Davor Ocelic's avatar
      Makefile improvements (#495) · 252518b5
      Davor Ocelic authored
      * Makefile improvements
      
      - Amber binary now depends on all its source files (src/**cr). If any file
        changes, just running 'make' will rebuild Amber. If no source files have
        changed, running 'make' will have no effect and will exit quickly.
        (One can always force make execution with 'make -B [target]', in case that
        forced recompilation is required.)
      
      - Structure of the Makefile is now such that adding more binaries later is
        easier
      
      - A couple "stylistic" improvements were made which don't affect previous
        behavior, but use more standard Makefile syntax and variables
      
      - When running 'make install', if previous copy or symlink doesn't exist
        (as is the case on a first install), 'rm' doesn't fail with critical error
      
      - 'make clean' now only deletes amber-specific files, not shards etc.
        This makes it easier/quicker to invoke 'make' afterwards and recompile
        amber without having to go through shards downloading again
      
      - 'make distclean' deletes all files and brings source tree close to original
        state as would be after git clone
      
      - Both variants of clean now only delete bin/amber and not bin/ as a whole,
        so they now correctly preserve 'ameba' binary which comes from shards and
        also lives in bin
      
      * Minor Makefile simplification
      252518b5
  4. 06 Jan, 2018 1 commit
    • Davor Ocelic's avatar
      Makefile improvements (#484) · 768fb6a9
      Davor Ocelic authored
      * Makefile improvements
      
      - Remove invocations of `pwd` where not necessary
      - Add invocations of `pwd` where necessary
      - Replace PREFIX?= with PREFIX=
      
      (The removal of ? in PREFIX?= makes this setting not susceptible to
      unintentionally taking the value from the environment variable.
      The correct way to set PREFIX is by doing 'make install
      PREFIX=/path'
      (as opposed to 'PREFIX=/path make install').)
      
      - Do not run link (which requires admin privileges) in default
        target (build)
      - On install, remove destination file before copying the binary
        (solves problem when a symlink already exists and copying fails)
      - Make build depend on lib/
      - Run shards only once, if lib/ doesn't exist
      - Better qualify $OUT_DIR
      - Add missing echo
      
      * Make link targets depend on build, just like install
      768fb6a9
  5. 20 Dec, 2017 5 commits
  6. 15 Sep, 2017 1 commit
  7. 10 Sep, 2017 1 commit
  8. 30 Aug, 2017 1 commit
    • Elias Perez's avatar
      [WIP] Remove amber directories from spec and src (#204) · a3d253a8
      Elias Perez authored
      * Remove amber directories from spec and src
      
      Moves the cmd directory outside of the amber directory and renames
      the core directory to Amber.
      
      Modules namespaces conventions should follow a directory structure
      where/when possible to keep the project cohesive and to understand
      where things are. For instance Amber::Controller should point to
      amber/controller directory.
      
      Currently there is a mismatched directory to module structure because
      of the core directory, amber/core/controller but the module is
      still Amber::Controller I tried/consider renaming the modules to include
      core, but it was just not right and the module namespacing became
      unnecessarily too long eg Amber::Core::Controller. For this reason,
      is better keep amber instead of core to maintain cohesion and logical
      file structure.
      
      This addresses the concern by
      
      - Removes the extra amber directory from src and spec directory.
      - Cleans and consolidates the specs
      - Organizes the spec directories
      - Changes namespace from Amber::CMD to Amber::CLI
      - Adds Server directory for server related classes/modules
      - It runs generated specs for the generated app
      - Cleaner source dir
      ```crystal
      -- src
        -- amber
        -- cli
        amber.cr
      ```
      
      What side effect does this change have
      
      - Since merging Amber::CLI with Amber CLI specs run with the Amber
      suite causing it to be slower.
      
      * minor bug fix
      
      * another path issue
      a3d253a8
  9. 27 Aug, 2017 1 commit
  10. 25 Aug, 2017 2 commits
  11. 11 Aug, 2017 1 commit
    • Jason Waldrip's avatar
      Docker and things (#71) · 5d719d70
      Jason Waldrip authored
      * update install method
      
      * Update dockerfiles
      
      * dont EVER include a license with something that could possibly not be OSS
      
      * Update Readme
      
      * fix specs
      5d719d70
  12. 30 Jul, 2017 1 commit
  13. 27 May, 2017 1 commit
  14. 22 May, 2017 1 commit
  15. 29 Jan, 2017 1 commit
  16. 31 Dec, 2016 1 commit
  17. 26 Dec, 2016 2 commits