wakie87 wrote on Tuesday, July 12, 2016:
Discussion continued from github: https://github.com/openemr/openemr/pull/150
Bradys last comment raises a lot of valid points:
As with bower and the front-end assets, I propose baby steps. This will at least be much simpler since there will only be 1 version of each package...
Recommend starting with placing plan on the wiki page:
http://www.open-emr.org/wiki/index.php/Composer
For an example, see the bower page:
http://www.open-emr.org/wiki/index.php/Bower
I see no reason to rock OpenEMR's world with separate repos and forced developer/tester composer education so early in the game. When implementing something, should also think of the downstream effect. For example, resources to:
1. convert githib openemr project to an organization
2. convert demo farm to build package
3. convert package builder script to build package
4. testing of packages will entirely fall on the person making the release
5. requires all developers(newbie, experienced, young, old) to use composer to have a working package. Thus need documentation for developers.
6. testers with minimal development experience whom test the master codebase frequently (ie. for example Arnab) will need to use composer to have a working package. Thus need documentation for non-developers.
Why not first just get composer in and start migrating packages to it (while tracking this on the wiki page). Then could always decide to go the build route in the future with very small amount of work, which would just involve deleting files and adding a phpmyadmin repo (ie. why are we making this decision now?).
If still wish to push for this so early in the game, then recommend discussing it on the forums.
-brady
Just wanting to continue the conversation here as I feel like we need to implement composer the right way from the very start.