diff options
author | Patrick Georgi <pgeorgi@google.com> | 2020-07-14 00:00:33 +0200 |
---|---|---|
committer | Patrick Georgi <pgeorgi@google.com> | 2020-08-31 20:23:22 +0000 |
commit | 4ad85ed156552e2bb61e28cdb51db0f2ba676bba (patch) | |
tree | f5642131a365e4f232a88fdbfa55ba9eeecc3267 /Documentation/community | |
parent | e4a7d9f69347314e7b3110a5f68921da768cb4e8 (diff) |
Documentation: Discuss how we use language
Change-Id: I44fa30af538c78760821401c8d3c52029d95b72b
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43420
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Diffstat (limited to 'Documentation/community')
-rw-r--r-- | Documentation/community/language_style.md | 136 |
1 files changed, 136 insertions, 0 deletions
diff --git a/Documentation/community/language_style.md b/Documentation/community/language_style.md new file mode 100644 index 0000000000..97f9601286 --- /dev/null +++ b/Documentation/community/language_style.md @@ -0,0 +1,136 @@ +# Language style + +Following our [Code of Conduct](code_of_conduct.md) the project aims to +be a space where people are considerate in natural language communication: + +There are terms in computing that were probably considered benign when +introduced but are uncomfortable to some. The project aims to de-emphasize +such terms in favor of alternatives that are at least as expressive - +but often manage to be even more descriptive. + +## Political Correctness + +A common thread in discussions was that the project merely follows some +fad, or that this is a "political correctness" measure, designed to please +one particular "team". While the project doesn't exist in a vacuum and +so there are outside influences on project members, the proposal wasn't +made with the purpose of demonstrating allegiance to any given cause - +except one: + +There are people who feel uncomfortable with some terms being used, +_especially_ when that use takes them out of their grave context +(e.g. slave when discussing slavery) and applies them to a rather benign +topic (e.g. coordination of multiple technical systems), taking away +the gravity of the term. + +That gets especially jarring when people aren't exposed to such terms +in abstract sociological discussions but when they stand for real issues +they encountered. + +When having to choose between using a well-established term that +affects people negatively who could otherwise contribute more happily +and undisturbed or an alternative just-as-good term that doesn't, the +decision should be simple. + +## Token gesture + +The other major point of contention is that such decisions are a token +gesture that doesn't change anything. It's true: No slave is freed +because coreboot rejects the use of the word. + +coreboot is ambitious enough as-is, in that the project offers +an alternative approach to firmware, sometimes against the vested +interests (and deep pockets) of the leaders of a multi-billion dollar +industry. Changing the preferred vocabulary isn't another attempt at +changing the world, it's one thing we do to try to make coreboot (and +coreboot only) a comfortable environment for everybody. + +## For everybody + +For everybody, but with a qualifier: We have certain community etiquette, +and we define some behavior we don't accept in our community, both +detailed in the Code of Conduct. + +Other than that, we're trying to accommodate people: The CoC lays out +that language should be interpreted as friendly by default, and to be +graceful in light of accidents. This also applies to the use of terms +that the project tries to avoid: The consequence of the use of such +terms (unless obviously employed to provoke a reaction - in that case, +please contact the arbitration team as outlined in the Code of Conduct) +should be a friendly reminder. The project is slow to sanction and that +won't change just because the wrong kind of words is used. + +## Interfacing with the world + +The project doesn't exist in a vacuum, and that also applies to the choice +of words made by other initiatives in low-level technology. When JEDEC +calls the participants of a SPI transaction "master" and "slave", there's +little we can do about that. We _could_ decide to use different terms, +but that wouldn't make things easier but harder, because such a deliberate +departure means that the original terms (and their original use) gain +lots of visibility every time (so there's no practical advantage) while +adding confusion, and therefore even more attention, to that situation. + +Sometimes there are abbreviations that can be used as substitutes, +and in that case the recommendation is to do that. + +As terms that we found to be best avoided are replaced in such +initiatives, we can follow up. Members of the community with leverage +in such organizations are encouraged to raise the concern there. + +## Dealing with uses + +There are existing uses in our documentation and code. When we decide to +retire a term that doesn't mean that everybody is supposed to stop doing +whatever they're doing and spend their time on purging terms. Instead, +ongoing development should look for alternatives (and so this could come +up in review). + +People can go through existing code and docs and sort out older instances, +and while that's encouraged it's no "stop the world" event. Changes +in flight in review may still be merged with such terms intact, but if +there's more work required for other reasons, we'd encourage moving away +from such terms. + +This document has a section on retired terms, presenting the rationale +as well as alternative terms that could be used instead. The main goal is +to be expressive: There's no point in just picking any alternative term, +choose something that explains the purpose well. + +As mentioned, missteps will happen. Point them out, but assume no ill +intent for as long as you can manage. + +## Discussing words to remove from active use + +There ought to be some process when terminology is brought up as a +negative to avoid. Do not to tell people that "they're feeling wrong" +when they have a negative reaction to certain terms, but also try to +avoid being offended for the sake of others. + +When bringing up a term, on the project's mailing list or, if you don't +feel safe doing that, by contacting the arbitration team, explain what's +wrong with the term and offer alternatives for uses within coreboot. + +With a term under discussion, see if there's particular value for us to +continue using the term (maybe in limited situations, like continuing +to use "slave" in SPI related code). + +Once the arbitration team considers the topic discussed completely and +found a consensus, it will present a decision in a leadership meeting. It +should explain why a term should or should not be used and in the latter +case offer alternatives. These decisions shall then be added to this +document. + +## Retired terminology + +### slave + +Replacing this term for something else had the highest approval rating +in early discussions, so it seems pretty universally considered a bad +choice and therefore should be avoided where possible. + +An exception is made where it's a term used in current standards and data +sheets: Trying to "hide" the term in such cases only puts a spotlight +on it every time code and data sheet are compared. + +Alternatives: subordinate, secondary, follower |