Difference between revisions of "Uno/API Deprecation"
From Apache OpenOffice Wiki
< Uno
(→If/How to Deprecate/Change/Remove UNO API) |
(→If/How to Deprecate/Change/Remove UNO API) |
||
| Line 1: | Line 1: | ||
= If/How to Deprecate/Change/Remove UNO API = | = If/How to Deprecate/Change/Remove UNO API = | ||
| − | Discussion has been going on for years what to do with known-stale/badly designed/un-used UNO IDL types. | + | Discussion has been going on for years what to do with known-stale/badly designed/un-used UNO IDL types. Here's an attempt to collect some ideas. |
Random ideas: | Random ideas: | ||
| − | * decide on rules when published types may be removed or changed | + | * decide on rules when published types may be removed or changed: |
** number of clients in OOo | ** number of clients in OOo | ||
| − | ** (estimated) use outside OOo project (i.e. people that don't read interface- | + | ** (estimated) use outside OOo project (i.e. people that don't read interface-announce) |
** how "bad" is the API - i.e. if everybody agrees it plainly sucks, remove anyway despite points above | ** how "bad" is the API - i.e. if everybody agrees it plainly sucks, remove anyway despite points above | ||
* deprecate in one feature release (add a suitable warning mechanism when using such API) | * deprecate in one feature release (add a suitable warning mechanism when using such API) | ||
* remove in the following feature release | * remove in the following feature release | ||
| − | * limit impact considerations | + | * limit impact considerations to non-ABI-dependent UNO bindings (i.e. the assumption is that c++ components break randomly anyway for every other release, so they shouldn't block API changes) |
| − | == Kill | + | == Kill file == |
List of problematic API that should be considered for deprecation/change/removal: | List of problematic API that should be considered for deprecation/change/removal: | ||
* css::document::XDocumentInfo [[User:mst]] | * css::document::XDocumentInfo [[User:mst]] | ||
* css::beans::StringPair - there's the generic beans::Pair<> type now [[User:thorsten]] | * css::beans::StringPair - there's the generic beans::Pair<> type now [[User:thorsten]] | ||
Revision as of 17:58, 30 January 2009
If/How to Deprecate/Change/Remove UNO API
Discussion has been going on for years what to do with known-stale/badly designed/un-used UNO IDL types. Here's an attempt to collect some ideas.
Random ideas:
- decide on rules when published types may be removed or changed:
- number of clients in OOo
- (estimated) use outside OOo project (i.e. people that don't read interface-announce)
- how "bad" is the API - i.e. if everybody agrees it plainly sucks, remove anyway despite points above
- deprecate in one feature release (add a suitable warning mechanism when using such API)
- remove in the following feature release
- limit impact considerations to non-ABI-dependent UNO bindings (i.e. the assumption is that c++ components break randomly anyway for every other release, so they shouldn't block API changes)
Kill file
List of problematic API that should be considered for deprecation/change/removal:
- css::document::XDocumentInfo User:mst
- css::beans::StringPair - there's the generic beans::Pair<> type now User:thorsten
