I have 3 separate database applications, two of which I have developed over several years and are fairly stable. Base is an excellent tool for these. One is a new application for which I have been doing quite a bit of development over a few weeks, including query design, form design (often with several nested subforms), Basic macros and dialogues. To cut this short, Base crashes 20 or 30 times a day whilst doing active development. I've never quite gathered a pattern to this, or at least not one strong enough to be worth spending time trying to replicate precisely. My impression is that most crashes happen when I close a window. Still, when I restart and Base recovers I usually lose a little work, but I've not had any long-lasting corruption. I accept this because I know the end result will be really useful, and (probably) stable, and Base is, after all, free for me to use. Nonetheless I thought I should still raise this to ask whether this is something many people encounter, or whether it is rare and so might be due to something specific on my platform. I apologise for not being very specific and not providing examples. This is partly because I haven't seen a pattern, and partly because I have been concentrating on finishing this stage of application development (as in: it crashed again, oh well, restart and carry on....).
Hi Tim, Did you try a new user profile? Ciao - Cor
(In reply to Cor Nouws from comment #1) > Hi Tim, > > Did you try a new user profile? > > Ciao - Cor No, I haven't. Is that likely to help?
(In reply to tim from comment #2) > No, I haven't. Is that likely to help? Sure - that's obviously the reason why I ask this :) https://wiki.documentfoundation.org/UserProfile
(In reply to Cor Nouws from comment #3) > (In reply to tim from comment #2) > > > No, I haven't. Is that likely to help? > > Sure - that's obviously the reason why I ask this :) > https://wiki.documentfoundation.org/UserProfile Sorry, silly of me to ask. I'll try it. I'm near the end of my development process now so expect fewer crashes anyway. I'd still be interested to know if my experience of frequent crashes when designing forms, macros etc is common or not.
(In reply to tim from comment #4) > Sorry, silly of me to ask. No problem ;) > I'll try it. I'm near the end of my development process now so expect fewer > crashes anyway. > > I'd still be interested to know if my experience of frequent crashes when > designing forms, macros etc is common or not. I've no idea. You could ask on user list or forum. But hard to tell if that will give a representative response.
I'll ask on the user forum. I was hoping that maybe people in the development team might have some views on this, assuming they have some test systems on which they test user development tools.
Created attachment 122180 [details] Test Database form I can now regularly reproduce one case where Base crashes. Tested on: Version: 5.1.0.2 Build ID: 1:5.1.0~rc2-0ubuntu1~trusty0 CPU Threads: 8; OS Version: Linux 4.2; UI Render: default; Locale: en-GB (en_GB.UTF-8) Open emptylms.odb Go to Forms Edit AlbumArtists form Change the shape of a push button The, open Tools, Macros, Organise Macros, Libreoffice Basic.. Select emptylms.odb, standard, select any macro module and Edit Close the AlbumArtists form with the x in the corner, and Save Close the odb with the x in the corner, and Save Base crashes and sometimes offers to restart now, sometimes not. This seems a relatively mild case of a crash, but during rapid development I can get dozens such each day, by no means all done this way - I haven't yet managed to reproduce others, but there are several that seem to occur when closing forms (not necessarily the whole odb). Sometimes it says there is a frame error on restart. This isn't a working odb, since the tables were imported from the true application using odbc and didn't import properly. That doesn't affect the crashes.
No repro for me with your test odb on Version: 5.1.3.2 Build ID: 644e4637d1d8544fd9f56425bd6cec110e49301b CPU Threads: 2; OS Version: Mac OS X 10.11.5; UI Render: default; Locale: fr-FR (fr.UTF-8) Loaded the ODB, entered form edit mode for the only form in the file, redimensioned the filter button, saved the change, then closed the form, went to Tools Macros Organize Basic etc, then opened various macro modules of the file in the Basic IDE - no crash.
@Tim : could this be yet another Ubuntu-LO-instability problem ?
(In reply to Alex Thurgood from comment #9) > @Tim : could this be yet another Ubuntu-LO-instability problem ? It seems not - it sill crashes every time using the precise sequence I mentioned earlier, and I sometimes get an ubuntu crash report. I'm now on ubuntu 16.04, LO 1.5.1.3-0ubuntu1
(In reply to Alex Thurgood from comment #8) > No repro for me with your test odb on > > Version: 5.1.3.2 > Build ID: 644e4637d1d8544fd9f56425bd6cec110e49301b > CPU Threads: 2; OS Version: Mac OS X 10.11.5; UI Render: default; > Locale: fr-FR (fr.UTF-8) > > Loaded the ODB, entered form edit mode for the only form in the file, > redimensioned the filter button, saved the change, then closed the form, > went to Tools Macros Organize Basic etc, then opened various macro modules > of the file in the Basic IDE - no crash. To get 5.1.3.2 to crash I did: - enter form edit mode for the only form in the file, - redimension the filter button (do not save or close now) - go to Tools Macros Organize Basic etc, then open a macro modules of the file in the Basic IDE and leave open - without minimising any windows, close the form using the window x button and when asked to save, say yes - without doing anything else, close the main database window using the window x button and when asked to save, say yes If the sequence is changed it doesn't always crash
Having done all this I retested in 5.1.4 rc1 today (the ubuntu pre-release), and it didn't crash. So I guess it has been fixed! Alex - could you double check 5.1.3.2 did crash as I said? I'll leave this for a few days before closing it.
I thought I should try a test under windows. In Windows 10, LO 5.1.2.2, it crashes exactly as I described, every time. I downloaded 5.1.4.1 to try that but unfortunately it doesn't run on my Windows 10 system. It complains that Java isn't installed, but it is, and is OK with IE. So thus far I can't prove the problem is universally fixed.
(In reply to tim from comment #13) > I thought I should try a test under windows. > > In Windows 10, LO 5.1.2.2, it crashes exactly as I described, every time. > > I downloaded 5.1.4.1 to try that but unfortunately it doesn't run on my > Windows 10 system. It complains that Java isn't installed, but it is, and > is OK with IE. So thus far I can't prove the problem is universally fixed. 2 things to check: 1) if you install LO 64 bits, you must have Java 64 bits (if it's LO 32 bits, you must have Java 32 bits) 2) rename your LO directory profile (see https://wiki.documentfoundation.org/UserProfile#Windows)
Created attachment 125481 [details] bt with symbols On pc Debian x86-64 with master sources updated today + following https://bugs.documentfoundation.org/show_bug.cgi?id=96877#c7, I could reproduce a crash.
(In reply to Julien Nabet from comment #14) > (In reply to tim from comment #13) > > I thought I should try a test under windows. > > > > In Windows 10, LO 5.1.2.2, it crashes exactly as I described, every time. > > > > I downloaded 5.1.4.1 to try that but unfortunately it doesn't run on my > > Windows 10 system. It complains that Java isn't installed, but it is, and > > is OK with IE. So thus far I can't prove the problem is universally fixed. > > 2 things to check: > 1) if you install LO 64 bits, you must have Java 64 bits > (if it's LO 32 bits, you must have Java 32 bits) > 2) rename your LO directory profile (see > https://wiki.documentfoundation.org/UserProfile#Windows) Ooops - that fixed it. Thanks. I guess I had the 32 bit LO installed previously without realising it.
I have now tried 5.1.4 RC 1 (64 bit) on Windows 10, and it doesn't crash for me there (or on ubuntu 16.04 as previously mentioned). If the very latest 'Masters' version still crashes (I can't easily test that) then something got better and then reverted.
The original reporter reports that the problem is fixed in 5.1.4 Juliens's master bt is the same bt as https://bugs.documentfoundation.org/show_bug.cgi?id=100140#c1 which is a 5-2/master only new problem. So I'll mark this as fixed as the original problem is apparently gone. While there is a new known problem in 5-2/master
I'll happily mark the Closed if someone else can confirm it was a problem prior to 5.1.4 and is now fixed there.
Tim: of course anybody can make a mistake but you can easily trust Caolán, he's a great core dev. (just give a look to https://cgit.freedesktop.org/libreoffice/core/log/?qt=author&q=mcnamara if you want to have an idea). You may also wait for final release of 5.1.4 and just use it to see.
(In reply to Julien Nabet from comment #20) > Tim: of course anybody can make a mistake but you can easily trust Caolán, > he's a great core dev. (just give a look to > https://cgit.freedesktop.org/libreoffice/core/log/?qt=author&q=mcnamara if > you want to have an idea). > You may also wait for final release of 5.1.4 and just use it to see. It's me I don't trust, not anyone else! I wouldn't dream of suggesting someone else was wrong in this type of thing. No one else has ever claimed to have seen the problem. that's all, and I thought confirmation would be the cautious way forward. So I simply meant that only I seemed to have had this problem and seen it had seemed to have gone away, even though not by any deliberate fix.
I wish I could try 5.2, but the fates are against me. The official Libreoffice collection of debs (i.e. not an ubuntu ppa) doesn't install properly on ubuntu 16.04 - there's some conflict in the installation that causes it to fail. There is a test 'snap' version which looks very interesting for the future, but it also has problems at present with some file accesses. I guess I will have to wait for an ubuntu ppa version.
I have now tested in 5.2.0.0 beta which wouldn't install (a deb conflict), and then when the installation problem was resolved it crashed and misbehaved in other ways. So I waited for 5.2.0.1. The problem has definitely been resolved in Version: 5.2.0.1 Build ID: fcbcb4963bda8633ba72bd2108ca1e802aad557d CPU Threads: 8; OS Version: Linux 4.4; UI Render: default; Locale: en-GB (en_GB.UTF-8) I'll close this. OK?