[haiku-bugs] Re: [Haiku] #15254: BWindow shouldn't dispatch messages on behalf of BView

  • From: "Haiku" <trac@xxxxxxxxxxxx>
  • To: undisclosed-recipients: ;
  • Date: Thu, 22 Aug 2019 15:17:32 -0000

#15254: BWindow shouldn't dispatch messages on behalf of BView
----------------------------------+----------------------------
   Reporter:  simonsouth          |      Owner:  nobody
       Type:  bug                 |     Status:  new
   Priority:  normal              |  Milestone:  Unscheduled
  Component:  Kits/Interface Kit  |    Version:  R1/Development
 Resolution:                      |   Keywords:
 Blocked By:                      |   Blocking:
Has a Patch:  0                   |   Platform:  All
----------------------------------+----------------------------

Comment (by simonsouth):

 (Posting this here as I am still without access to Gerrit; see #14884.)

 Following on from [https://review.haiku-os.org/c/haiku/+/1723 the
 discussion on Gerrit]:

 To be clear, there isn't a specific problem this change solves for me.
 What it does do is lead to code that is simpler, more efficient and more
 general in nature. It also brings Haiku in line with how Be's
 documentation (as well as common sense, in my opinion) says the API should
 behave: I would argue the
 [https://en.wikipedia.org/wiki/Principle_of_least_astonishment principle
 of least astonishment] is violated by a message-passing system that stops
 a message's flow halfway through and starts invoking hard-coded handler
 methods instead.

 It seems unlikely the patch will lead to any significant regressions; for
 a week now I've been running a Haiku image with the patch applied and
 haven't noticed any issues. However, if anything ''does'' break, I'd
 rather fix it all now while Haiku is still young, before design quirks
 like these have had time to ossify.

-- 
Ticket URL: <https://dev.haiku-os.org/ticket/15254#comment:11>
Haiku <https://dev.haiku-os.org>
The Haiku operating system.

Other related posts: