[CLOSED] Editor loses position when switching from file to file #5756
Comments
|
Oh, and same thing happens for undo. It seems that most of the time when I change alt+tab to fe. browser and them come back the undo history is gone. |
|
I do see the scroll position get lost occasionally (jumping back to the top of the file) when switching files, but never simply from Alt+Tabbing in & out of Brackets. And I've never heard of the Undo history disappearing -- that's quite serious. Is the file still marked as "dirty" (unsaved changes -- indicated by a dot/bullet next to the file name), but Undo does nothing? Are you sure the editor still has keyboard focus when you're attempting to undo? Fwiw, if all you're seeing is the scroll pos getting lost, try tapping the left/right arrows: in the bug I've seen, the cursor position is still correct, so moving the cursor even 1 char will cause the scroll position to jump back down to the correct place. |
|
The undo history seem to disappear the same way, alt-tabbing away. Pretty much every time I save and alt-tab to browser. Although, now that I think of it, I might have a corrupted install (even though I've reinstalled it) as fe. key remappings never seem to stick (they always reset back to normal after exiting brackets). I think I'll do a clean install (uninstall, remove files & folders) and then try things out with vanilla Brackets. |
|
Ok. I did clean install no Brackets and running with no extensions installed. The undo history just disappeared on me again. Saved, alt-tabbed to browser, refresh, alt-tab back and no undo history. Another thing that might have something to do with this - For some reason Brackets keeps telling me every now and then that the contents of the file has been changed and do I want to load the latest from disk. No other editor has done this kind of behavior before and I'm quite sure nothing has changed in the file, unless some rogue process in the virtual Ubuntu has done something but it's a clean install, so I doubt it. Console has one error, though since I didn't check it before I'm not sure if it's related to the issue Oh and no, it wasn't "dirty", as I had just saved it. Also, yes, I'm sure the focus was on the editor, was moving the caret around. :) |
|
Is it possible there's some flaw in your (unusual sounding) SSH-backed VM filesystem setup that could cause file IO operations to report success but only later, after some async delay, begin reflecting that update in read/ Just to make sure it's not anything specific to your project setup, could you try editing the basic "Getting Started" project that ships with Brackets and see if the same problem occurs there? Btw, are you running the official Sprint 35 build from download.brackets.io, or some other distro?... or building from git directly? |
|
Yes but I never reload it from the disk, I always use the one I have in the editor. Oh, and the undo history does disappear even without that happening, can happen just with alt-tabbing. I have Eclipse and ConTEXT as a backup editors, and those haven't had the same issue a single time so far, so it's either those having better error handling on that regard or Brackets is doing something differently. Well, I can try to test that, though as it doesn't happen every time I don't think it's an issue with just random edits on the getting started project. Though, I have other projects that aren't run in Linux environment (meaning, no SSH mounting), so I can report on those later on as I can't remember for sure if it happened with those as well, though I'm thinking yes. Yeah, official Sprint 35, with clean install and no extensions. |
|
Also, just to clarify your environment... is this a correct description?
|
|
Two other things we could we could try for diagnostics:
|
|
Hmm, interestingly I found the following note in the WriteFile docs:
"Unbuffered I/O" apparently refers to But then I'm getting conflicting information from other MSDN docs:
This sounds both scarier on the one hand (implying could get incorrect timestamps on any filesystem, not just remote ones... maybe), and better on the other hand (offering two other solutions that are both simpler than (See more docs about
CC Seems like we might at least want to try setting up a network share and Win machine that can reproduce this behavior, for testing... |
|
We should also test this on a Mac with an SSH share but it may be easier for those without Windows expertise to dive into and debug if we can repro it. (http://www.alirezabagheri.com/blog/?p=67) I'm not sure how |
|
Hey. Unfortunately I Haven't had time to properly experiment outside the SSH mount on that computer. This is the exact environment on that case:
|
|
Finally had possibility to work with the laptop in question (with the SSH mounts). No issues outside SSH mount, undo history & file position seems to be remembered correctly. |
|
|
|
Hey. Unfortunately I'm not running that environment anymore as that project Mikko Tikkanen ResponsiveAds Zonear Ltd. Mint Usability 2014-03-18 7:15 GMT+02:00 Peter Flynn notifications@github.com:
|
|
Closing since there's no way to gather further data and we're unable to reproduce this on our end.
|
Friday Dec 27, 2013 at 14:54 GMT
Originally opened as adobe/brackets#6326
Editor loses the current position from time to time and reverts to beginning of the file. Quite a pain with big files.
As a side note, this seems to happen on a setup where Brackets is running on a Windows machine with virtual Ubuntu, which is mounted as a network drive, wherein all the code lives. So, edits happen through the SSH mapped network drive.
Could the editor be using too unspecific data for storing the editor positions? (so outside processes touch the files, it looks like a new file for the editor and the position is lost?)
The text was updated successfully, but these errors were encountered: