On 15-10-31 02:08 AM, Thomas Martitz wrote:
Am 31. Oktober 2015 02:52:01 MEZ, schrieb Matthew Brush notifications@github.com:
[...] so unless there are actual problems I don't see an evidence to
keep the off default.
I mentioned a number of issues with it in the mailing list thread linked from the PR you made to disable it by default.
I don't think we should cram this in just because of string freeze, especially with the open issues, the UI issues, and that we will be releasing again in a few months. I'm not opposed to the feature, I'm opposed to the implementation (re-using an existing feature to do something else, something unexpected of the existing feature, as well as the infobar without a checkbox that can be hidden from the user without ever seeing it).
Reply to this email directly or view it on GitHub: https://github.com/geany/geany/pull/672#issuecomment-152685983
Carrying this discussion to the ML.
Can you elaborate on the issues you see?
Mostly the stuff I mentioned in the previous mailing list thread on this subject. The semantics of the Reload feature have changed and no longer does reload mean "drop everything about this file and load it again" it means, "store another copy of this file in RAM and load it again". I feel like it's not a good idea to re-use the File->Reload feature like this, and under not uncommon scenarios, Geany is no longer "lightweight" with respect to RAM usage.
To give an example, if someone opened a big log file and Reloads it, or if they have many files from version control open and switch branches, every reload will "leak" the size of each document worth of memory, making Geany use much more RAM than before. Scintilla's undo mechanism just isn't designed to efficiently handle the case of the entire document being replaced. The only workaround to regain all the memory wasted by Geany is to close all the files and start over.
I think the new preference is fine, I just don't think it ought to be enabled by default, or else it ought to be associated with a new (Edit, not File) action altogether as opposed to changing the semantics of the existing feature as everyone is used to.
As to the UI. It's important that we get any UI for 1.26, therefore we had to manage before string freeze. [...]
Why is it important to rush it in before 1.26? One of the points of the accelerated release cycle is so that people would stop wanting to cram stuff in right before release. It's only a few months until next release, so there's no need to keep trying to rush in controversial stuff.
Cheers, Matthew Brush