#general
- w
witty-crayon-22786
09/16/2015, 5:20 PMno. - b
billowy-alarm-80641
09/16/2015, 5:20 PMsure, if it's a private API - w
witty-crayon-22786
09/16/2015, 5:20 PMright. it is. - m
mammoth-pharmacist-54215
09/16/2015, 5:20 PMits not truly internal if you have the ability to interact with and change settings in pants.ini and command line args. - w
witty-crayon-22786
09/16/2015, 5:20 PM"internal" - b
billowy-alarm-80641
09/16/2015, 5:20 PM+1 - h
happy-kitchen-89482
09/16/2015, 5:20 PMPants allows you to add arbitrary jvm arguments to any invocation, and it's possible that some install base was setting that. but in practice no one is. - b
billowy-alarm-80641
09/16/2015, 5:20 PMthe line between private / public is really fuzzy w/ pants - h
happy-kitchen-89482
09/16/2015, 5:21 PMWe can't make the entire surface area of every internal JVM invocation part of pants's external API. - h
happy-kitchen-89482
09/16/2015, 5:21 PMWe could lock down what we allow people to set. - h
happy-kitchen-89482
09/16/2015, 5:21 PMI.e., we could prevent all external tinkering with inernally-generated command lines - h
happy-kitchen-89482
09/16/2015, 5:21 PMBut that might be a little too restrictive at this point. - w
witty-crayon-22786
09/16/2015, 5:21 PMimo, in this case, unless pants exposes a native pants.ini option for the zinc flag, we should consider it to be private/experimental. - h
happy-kitchen-89482
09/16/2015, 5:22 PM=stuhood - m
mammoth-pharmacist-54215
09/16/2015, 5:22 PM@happy-kitchen-89482: My point is that this is an OSS project, and if we want pants to be generally useful (not just for a select few companies) we should have good practices around deprecation and function changes that may break people. - f
fancy-queen-20734
09/16/2015, 5:22 PMThe problem I have with it is that zinc has supported it since 2012. If there are zinc users who are/have migrated to pants that use the flag they might be surprised when it doesn’t exist at all - f
fancy-queen-20734
09/16/2015, 5:23 PMSo I think we should at least note removing it in our Zinc’s changelog - e
enough-analyst-54434
09/16/2015, 5:23 PM@mammoth-pharmacist-54215: I think we do, @ancient-stone-50795 got us on the path a while back now - we generally have excellent deprecation protocol save for CHANGELOG discoverability - e
enough-analyst-54434
09/16/2015, 5:23 PMThis is a very different case - h
happy-kitchen-89482
09/16/2015, 5:24 PMI understand your concern in principle, and if this were a pants option, part of the "expected" pants interface, we would do a proper deprecation cycle. But it's just not practical to consider every JVM option that we technically allow a repo to set as something to undergo deprecation. - m
mammoth-pharmacist-54215
09/16/2015, 5:24 PMMy point is that as an end user I shouldn’t have to worry about implementation details of pants… if I can set something as a command line option or an option in pants, its not an “internal” feature - h
happy-kitchen-89482
09/16/2015, 5:24 PMNot really - w
witty-crayon-22786
09/16/2015, 5:24 PM@mammoth-pharmacist-54215: end users don't set this flag. repo owners do. - h
happy-kitchen-89482
09/16/2015, 5:25 PMand even repo owners can only set it if they've taken the trouble to learn zinc internals... - w
witty-crayon-22786
09/16/2015, 5:25 PMright. - h
happy-kitchen-89482
09/16/2015, 5:25 PMotherwise they wouldn't know it exists - e
enough-analyst-54434
09/16/2015, 5:25 PMIts a flag 3 people in the world know about - m
mammoth-pharmacist-54215
09/16/2015, 5:25 PMrepo owners are “end users” from the perspective of pants devs… they are our users in the same way the web admins are apache’s end users. - f
fancy-queen-20734
09/16/2015, 5:25 PMor, called zinc -help at some point - w
witty-crayon-22786
09/16/2015, 5:25 PMwe have to draw a line.