You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Next thought, does it start out looking like that, or is the corruption triggered by something? eg it starts out looking ok, but corruption happens when making the window bigger (for example)
Trying to make a mental picture here of what the cause could be. 😄
Yeah, it's a bit of a head scratcher
Some data points:
1) Yes, this is under VirtualBox ( Version 5.1.12 r112440 (Qt5.5.1) )
2) The screen does start out that way, at least for that particular database. When I load others (which are also included in that link I added. See "/data/databases/" ) they look fine. Is the problem database
too big, or the schema too long or something along those lines?
3) With the problem database, it is just that one column "schema" in the one pane "Database Structure" which is messed up. The "DB Schema" panel on the right side, which looks like exactly the same data to me, is not corrupted in the same way. See attached screenshot
4) This is all using the current version as of tonight from the PPA. Here's the "about" info:
Version 3.9.99
Qt Version 5.6.1
SQLite Version 3.14.1
@jlmcgraw Has that link changed? Just tried it and got a 404. 😉
I'm kind of thinking it's likely a bug in VirtualBox that's being triggered by some weird edge case.
Random things to try, that might start to show the shape of the problem:
In the VirtualBox settings, see if changing graphics related things helps. eg Try different video memory sizes
Also in VirtualBox, do you have the Guest Addons loaded for the VM? Not sure if it'll make a difference but try changing this (if possible for MATE 16.10)
On a different tangent, are you ok with compiling things from source code? It'd be interesting to see if using a version compiled locally in the VM itself has the same problem. Just in case it's caused by some weird mismatch in (for example) the Qt libraries on your VM vs the system it was compiled on.
I guess it comes down to... how much effort are you willing to spend to investigate? 😀
Try this link: https://www.dropbox.com/s/fk1uup3zx7sb8r0/data.tar.xz?dl=0
I'm certainly willing to do some work on it. I do have the guest additions installed and I'll try adjusting a few settings related to graphics.
I can also work on compiling from source if you think there's a difference between it and what is in the PPA.
FWIW, this is the only application I see corruption in and it's only in that one pane (and the one database, of the few I've tried sqlitebrowser with, though of course that may be a red herring)
@innermous Yeah, it's more likely a Qt issue related to some weird virtualbox bug. It doesn't mean we should ignore it though, as it is obviously affecting DB4S.
@jlmcgraw Definitely try compiling DB4S (our abbreviation for sqlitebrowser) on the VM. I'm hoping 😄 it's a weird interaction between VirtualBox and the Qt libraries of the system it was compiled on. If that's the case, then using a version built on the VM has a reasonable chance of working.
@jlmcgraw It's very late in my TZ, so I'm hitting the sack in the next few minutes. I'll take a look at this issue again tomorrow and see if you've made any progress.
Btw, the new download link worked fine. Haven't investigated the data set yet (a tomorrow task). 😄
No problem, thanks for all of the help so far.
I've done the following and I'm still seeing the same issue
1. Clone repository from github
2. cmake .
This produced some messages and failed, which lead me to issue #635 and then to installing some qt5 related stuff
3. sudo apt install qt5-default qttools5-dev-tools
4. cmake .
This produced some messages but seemed to complete (see attached)
5. make
This succeeded (see attached) and produced an executable
6. execute ./sqlitebrowser
Same corruption issue
7. sudo make install && sqlitebrowser
Same corruption issue
Another note: I tried rearranging the columns so "type" comes after "schema" and it is just the "schema" column that is corrupted
http://pastebin.com/Rbm8HwNT
I've discovered something interesting: I deleted the APT_RWY and APT_APT tables and the corruption is gone!
What is it about those tables? The length of the schema?
That's fine with me. Did you want to look deeper into what interaction is triggering this issue since this is an experience users may receive until the distributions update their qt libraries? Or did I manage to stumble across some unique screwball corner case that isn't worth bothering with?
Well, (to me) it seems like a Qt bug that there's not much we can do about until the Qt library in Ubuntu is updated. It's probably going to be triggered by all kinds of weird stuff, and isn't really worth the time to investigate more as we're already kind of overloaded with things.
Note - It's not a complete loss, people will be able to find the details in this issue here with some simple searching. 😄
@chrisjlocke Good point. I haven't created any real reference data sets on the dev1 server yet mainly because I've been working on the visualisation stuff.
But yeah, it doesn't hurt to upload it there to make things easier. 😉
It's over 200MB in size... that's the first one of largish size I've tried btw. Kind of surprised it uploaded and downloaded without issues, though it did. Upload was a bit slow, but that's probably because it's not a beefy server at the moment.
Did a binary compare of the database file downloaded from DBHub, to the original... it's bit perfect. Yay. 😉
On Jan 6, 2017, at 9:09 AM, Chris Locke ***@***.***> wrote:
Always good to test 'real data'. ;)
200 MB for a database (uncompressed) is quite minimal really....
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the thread.
I was able to reproduce this under Ubuntu 18.04 with the database provided in #1965. Qt version is 5.9.5. It is only reproduced if the QT_QPA_PLATFORMTHEME environment value is set and with value gtk2. This isn't the Ubuntu default, I set it in my profile, so that DB4S looks the same as other applications in the desktop (default dark theme), otherwise it uses a clear theme, with which this problem is not reproduced. I guess that using QT_QPA_PLATFORMTHEME=gtk2 is buggy and Ubuntu does not use it, although it has the effect of Qt applications having a different look than other desktop applications.
I don't open this issue again, because it seems a pure Qt bug, but I felt like it was a good idea to record this information here.
It is possible to fix the display problem using qt5c and setting a font with size 12, but there is still a performance problem and the CPU usage of sqlitebrowser and Xorg are very high.
Activity
justinclift commentedon Jan 5, 2017
Well, that's different. 😉
Doing some searching online shows various issues with screen corruption in VirtualBox, and also some with AMD cards.
Does either of that apply to you? 😄
justinclift commentedon Jan 5, 2017
Hmmm, also wondering if it's maybe a Qt thing. Hmmm....
justinclift commentedon Jan 5, 2017
Next thought, does it start out looking like that, or is the corruption triggered by something? eg it starts out looking ok, but corruption happens when making the window bigger (for example)
Trying to make a mental picture here of what the cause could be. 😄
jlmcgraw commentedon Jan 5, 2017
vtronko commentedon Jan 5, 2017
I don't think it's DB B related issue, seriously.
justinclift commentedon Jan 5, 2017
@jlmcgraw Has that link changed? Just tried it and got a 404. 😉
I'm kind of thinking it's likely a bug in VirtualBox that's being triggered by some weird edge case.
Random things to try, that might start to show the shape of the problem:
On a different tangent, are you ok with compiling things from source code? It'd be interesting to see if using a version compiled locally in the VM itself has the same problem. Just in case it's caused by some weird mismatch in (for example) the Qt libraries on your VM vs the system it was compiled on.
I guess it comes down to... how much effort are you willing to spend to investigate? 😀
jlmcgraw commentedon Jan 5, 2017
justinclift commentedon Jan 5, 2017
@innermous Yeah, it's more likely a Qt issue related to some weird virtualbox bug. It doesn't mean we should ignore it though, as it is obviously affecting DB4S.
justinclift commentedon Jan 5, 2017
@jlmcgraw Definitely try compiling DB4S (our abbreviation for sqlitebrowser) on the VM. I'm hoping 😄 it's a weird interaction between VirtualBox and the Qt libraries of the system it was compiled on. If that's the case, then using a version built on the VM has a reasonable chance of working.
justinclift commentedon Jan 5, 2017
@jlmcgraw It's very late in my TZ, so I'm hitting the sack in the next few minutes. I'll take a look at this issue again tomorrow and see if you've made any progress.
Btw, the new download link worked fine. Haven't investigated the data set yet (a tomorrow task). 😄
jlmcgraw commentedon Jan 5, 2017
jlmcgraw commentedon Jan 5, 2017
vtronko commentedon Jan 5, 2017
Use hastebin or alternatives for such long outputs next time, please :)
jlmcgraw commentedon Jan 5, 2017
7 remaining items
jlmcgraw commentedon Jan 6, 2017
justinclift commentedon Jan 6, 2017
Well, (to me) it seems like a Qt bug that there's not much we can do about until the Qt library in Ubuntu is updated. It's probably going to be triggered by all kinds of weird stuff, and isn't really worth the time to investigate more as we're already kind of overloaded with things.
Note - It's not a complete loss, people will be able to find the details in this issue here with some simple searching. 😄
justinclift commentedon Jan 6, 2017
@jlmcgraw With the spatial56day.db database, it seems good for testing our SQLite parser with (an area of things we're working on).
Is it ok with you if we hang onto it and use it for development stuff?
jlmcgraw commentedon Jan 6, 2017
justinclift commentedon Jan 6, 2017
Awesome @jlmcgraw, thanks. 😄
@MKleusberg The archive here has a database in it (/databases/spatial56.db) that seems useful for your parser testing:
https://www.dropbox.com/s/fk1uup3zx7sb8r0/data.tar.xz?dl=0
chrisjlocke commentedon Jan 6, 2017
...and that hasn't been stored on dbhub.io because..... ;)
vtronko commentedon Jan 6, 2017
@chrisjlocke nice catch
justinclift commentedon Jan 6, 2017
@chrisjlocke Good point. I haven't created any real reference data sets on the dev1 server yet mainly because I've been working on the visualisation stuff.
But yeah, it doesn't hurt to upload it there to make things easier. 😉
justinclift commentedon Jan 6, 2017
@chrisjlocke It's on DBHub now:
https://dev1.dbhub.io/justinclift/spatial56day.db 😄
It's over 200MB in size... that's the first one of largish size I've tried btw. Kind of surprised it uploaded and downloaded without issues, though it did. Upload was a bit slow, but that's probably because it's not a beefy server at the moment.
Did a binary compare of the database file downloaded from DBHub, to the original... it's bit perfect. Yay. 😉
chrisjlocke commentedon Jan 6, 2017
Always good to test 'real data'. ;)
200 MB for a database (uncompressed) is quite minimal really....
jlmcgraw commentedon Jan 6, 2017
justinclift commentedon Jan 6, 2017
Heh Heh Heh, good points. 😄
mgrojo commentedon Aug 13, 2019
I was able to reproduce this under Ubuntu 18.04 with the database provided in #1965. Qt version is 5.9.5. It is only reproduced if the QT_QPA_PLATFORMTHEME environment value is set and with value gtk2. This isn't the Ubuntu default, I set it in my profile, so that DB4S looks the same as other applications in the desktop (default dark theme), otherwise it uses a clear theme, with which this problem is not reproduced. I guess that using QT_QPA_PLATFORMTHEME=gtk2 is buggy and Ubuntu does not use it, although it has the effect of Qt applications having a different look than other desktop applications.
I don't open this issue again, because it seems a pure Qt bug, but I felt like it was a good idea to record this information here.
mgrojo commentedon Aug 13, 2019
It is possible to fix the display problem using qt5c and setting a font with size 12, but there is still a performance problem and the CPU usage of sqlitebrowser and Xorg are very high.