Description: I will attach a document that is affected by this bug. Follow the Steps to Reproduce. It consistently crashes my system (Xubuntu 18.04LTS with XFCE 4.14) Steps to Reproduce: 1. Open the attached Writer document 2. Menu Insert > Field > More Fields 3. Tab Variables, Type: User Field, Select py3o.d.lender (or any field that is not currently used in the document) 4. Click the Button to delete the field Actual Results: crash Expected Results: the field is removed Reproducible: Always User Profile Reset: No Additional Info:
Created attachment 155123 [details] test case
Repro LO 6.4+ in Windows, so All. No repro 6.2, regression from 6.3. There's also crash report. Y, if you see the link, please write it.
Thanks for the quick reply, Timur. I do not understand what is the crash report? where would I see the link? where should I write it?
Crash report is a notification that LO crashed, if you get it you would see a link on it. If you're asking, you maybe didn't get it. Never mind. Nprmally we write it on the right side in Crash report field, but here in Comment would be ok. Anyway bibisect is more useful.
Created attachment 155132 [details] bt with debug symbols On pc Debian x86-64 with master sources updated today, I could reproduce the crash.
I proposed a patch here: https://gerrit.libreoffice.org/#/c/81114/ but not sure it's the right fix (see comment in gerrit)
Julien Nabet committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/09a3762fe0fc4c4815c842098094082bf1b90de6 tdf#128232: fix crash when trying to delete unused User Field It will be available in 6.4.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback.
Patch for 6.3 is on gerrit review here: https://gerrit.libreoffice.org/#/c/81120/
Julien Nabet committed a patch related to this issue. It has been pushed to "libreoffice-6-3": https://git.libreoffice.org/core/commit/9b1f906ea2aa946e9d6245baacd4937233bb58c6 tdf#128232: fix crash when trying to delete unused User Field It will be available in 6.3.4. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback.
Wow, you guys are amazing, thanks a heap! I would like to test the fix and report back. HOWEVER reporting this bug revealed something that is of greater concern to me and I will have to fix that first. Maybe with your help. Because you as an organization (Document Foundation / LibreOffice) have to understand that there is a deep fault in the way the app collects data, even if it is for a good purpose, and fix data collection accordingly. Thanks Timur for making me aware of a possible "crash report." While a crash report can be a good thing, it can also be a data leak. Bad. I was not asked for permission before LibreOffice called home at crashreport.libreoffice.org. There are many good reasons for this kind of telemetry, and I have no doubt that LibreOffice means well. HOWEVER, there is an overriding reason to ask for user consent first. I am a lawyer. I could have been working on highly confidential information, What if a memory dump is included in the crash report, and that confidential information is in the memory dump? I will not download a piece of software from https://dev-builds.libreoffice.org/daily/ when I know I cannot trust it to do the Right Thing. The fact that the competition is worse ("implied" consent buried into Microsoft's licensing terms) is no justification for not doing the Right Thing. I am happy to contribute in the form of a bug report; and even a *controlled* dump and backtrace; and even continuing on to using bisecting builds from repository until the faulty commit is identified; and even attempting to contribute a patch. HOWEVER, I am not willing to compromise on my professional duty of confidentiality. Nor am I willing to compromise on the right of each user to their own data confidentiality. Please point me to (1) a preference to prevent the collection and transmission of crash reports (2) an exact description of how crash reports are generated and what data is gathered and how the data is processed, transmitted, stored, destroyed (3) instructions how to build LibreOffice from source for my own machine
Y: I understand your point of view, I just think you should submit a new bugtracker (enhancement type). Michael: thought you might be interested in this one since it concerns legal things.
Thanks, Julien. I agree with you about a new bugtracker and submitted Bug 128255. My intention was to submit one when I have enough knowledge of what I write about. I do not know: WHAT, HOW, WHEN is collected. All I know is that there was an attempt to call home without me having enough information to consent to it. But then, I recalled an important principle of successful open source: release early, release frequently. So I "released" my thinking there. Thank you also for connecting with Michael / legal. I am more than willing to volunteer the time necessary to work things out and help make LibreOffice better for the public good. No intention to take the issue of informed consent to telemetry as a legal battle. For me personally, the solution if LibreOffice does not do the Right Thing is simple: fork the source code, maintain patches to remove the offending code, roll my own. But it is preferable to contribute user time and effort upstream. I can reasonably prove that some fundamental choices made by the IT industry as a whole in the last 10-15 years are bad. Many of them also cause harm. Telemetry without informed consent being the one in cause here and now. I am also of the opinion that the forum to solve this is anti-trust, not product liability. And I am of the opinion that a software publisher can do the Right Thing from the get go and avoid the whole controversy: no data collection without express and informed user consent. My request for instructions to build LibreOffice on my own machine still stands. I had to recently upgrade after my old PC died, and my glorified typewriter now has more CPU power than it needs. I am almost finish setting it up as "homelab", with my desktop being one of the virtual machines. Once my setup is robust, up and running, maintaining a VM to build and test such an important app as LibreOffice is something that will cost me little effort and that I will gladly do, as a little token of appreciation, and because sharing is the Right Thing to do.
Thanks for the pointer Julien. Y - we take people's privacy seriously, of course. > I was not asked for permission before LibreOffice called home at > crashreport.libreoffice.org. There are many good reasons for this kind of > telemetry, and I have no doubt that LibreOffice means well. That is un-expected; I would imagine that you would get a dialog (something like the attached) at the next restart that prompts you to send the crash-report. People tend to have a habit of clicking through those dialogs - if you're at all concerned don't send the report > HOWEVER, there is an overriding reason to ask for user consent first. > I am a lawyer. I could have been working on highly confidential > information, What if a memory dump is included in the crash report, > and that confidential information is in the memory dump? I would not expect to see any significant document data inside the breakpad / stack. > The fact that the competition is worse ("implied" consent buried into > Microsoft's licensing terms) is no justification for not doing the > Right Thing. Agreed. Then again contributing crash reports is extremely helpful for the project. > (1) a preference to prevent the collection and transmission of > crash reports The setting is: officecfg/registry/schema/org/openoffice/Office/Common.xcs- <prop oor:name="CrashReport" oor:type="xs:boolean" oor:nillable="false"> Tools->Options->Advanced->Open Expert search for 'CrashReport' > (2) an exact description of how crash reports are generated and what > data is gathered and how the data is processed, transmitted, stored, > destroyed I'm afraid you'll have to dig this out yourself, although Markus may have some more details easily to hand. > (3) instructions how to build LibreOffice from source for my own machine That's easy - instructions are easy to find from Google - but I'll post them here for you this once: https://wiki.documentfoundation.org/Development/How_to_build There is a configure option for breakpad support: --enable-breakpad which defaults to off (it seems). Poke at: ./desktop/Library_crashreport.mk ./desktop/source/app/crashreport.cxx and git grep -5 ENABLE_BREAKPAD I hope that helps =)
Created attachment 155155 [details] a picture I attach a snapshot of the dialog from glade NB. several strings are shown at once (for ease of translation reasons) that are from later / conditional steps in the flow through the dialog.
Created attachment 155156 [details] glade dialog picture - note the associated comment. I attach a snapshot of the dialog from glade NB. several strings are shown at once (for ease of translation reasons) that are from later / conditional steps in the flow through the dialog.
Thanks for your time and explanations, Michael. I never saw the Crash Report Dialog. To compare, I fired up a Windows VM, installed LO's from the LO website (6.3.2.2) and triggered the bug. No such dialog there either. > Tools->Options->Advanced->Open Expert search for 'CrashReport' Empty. Both Ubuntu and Windows versions. IMHO important enough to be in ->General, not in ->Advanced. And with a proper explanation in very close distance (mouseover/link?) > contributing crash reports is extremely helpful for the project. Agree. This is why I try contribute when I can; and this is why I am advocating for making the process a respectful and encouraging one. Empowerment encourages. Being told that "we know what is good for you and the rest is visual noise" is discouraging. Further remarks on Bug 128255. The key words to *contributing* are KNOWINGLY and VOLUNTARILY. > I would not expect to see any significant document data inside the breakpad > / stack. Happy to rely on your expectations in the general context; however, when client information is in the documents being processed, I need reliable certainty, not "expectations." Nothing personal. Better be safe than sorry. > > (2) an exact description of how crash reports are generated and what > > data is gathered and how the data is processed, transmitted, stored, > > destroyed > > I'm afraid you'll have to dig this out yourself, although Markus may have > some more details easily to hand. Happy to dig this out myself and to contribute a text for public consumption. You have given me enough hints where to start. Not sure when I can fit this into my time planning. In the meanwhile, I do not have the necessary information to consent to a crash report and crashreport.libreoffice.org is now blacklisted on my network. Sometimes, the path of least resistance is the temporary solution that becomes permanent. > > (3) instructions how to build LibreOffice from source for my own machine > > That's easy - instructions are easy to find from Google - but I'll post them > here for you this once: > https://wiki.documentfoundation.org/Development/How_to_build You are assuming that I am using Google. A stereotypical example of what has gone wrong in the IT industry. Anyway, thank you very much for your help. Much appreciated.
Verified in Version: 6.4.0.0.alpha1+ Build ID: de4839e66d3d195315729b95cc144cdab96b6e74 CPU threads: 4; OS: Linux 4.15; UI render: default; VCL: gtk3; Locale: ca-ES (ca_ES.UTF-8); UI-Language: en-US Calc: threaded @Julien Nabet, thanks for fixing this issue!
> I never saw the Crash Report Dialog. Then surely you didn't file the crash report we are talking about. I imagine someone else did. Timur - do you have a link to the crash report in question - it might help clarify things. Beyond that - if people have genuine legal concerns, bugzilla is not the place to air them - please contact legal@documentfoundation.org and please be aware of the -extraordinary- expense in engineering time it takes to prove a negative. We believe we are doing this right, having concrete, technical evidence that we are not would be extremely helpful - again on that list. As for other questions about how we handle bugs - those probably all belong on the QA list or in a QA call, no doubt we can improve our processes: but from my perspective getting bugs closed and keeping the set of things we focus on reasonably small is really important.
(In reply to Michael Meeks from comment #18) > > I never saw the Crash Report Dialog. > > Then surely you didn't file the crash report we are talking about. I imagine > someone else did. Timur - do you have a link to the crash report in question > - it might help clarify things. I just added the link to the see also list
Thanks Xisco - so - Y - to see if there is a report from you filed in error - we are going to need your exact version to filter the reports - and also a rough day/time it crashed for you. Also if you see a report from today eg. https://crashreport.libreoffice.org/stats/crash_details/2927b9eb-3e32-4963-b8c1-9f7e1872687d You can click the 'Raw' tab and see the raw textual content of the report in case that is interesting. Thanks.
Julien Nabet committed a patch related to this issue. It has been pushed to "libreoffice-6-3-3": https://git.libreoffice.org/core/commit/53e0240109b22db485b4b1212cf63cc8fa4de2be tdf#128232: fix crash when trying to delete unused User Field It will be available in 6.3.3. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback.
Although it's fixed, here is the bibisect. ceb2a932e352b0c2af9fbaa2ad62ac9f1b3069eb is the first bad commit commit ceb2a932e352b0c2af9fbaa2ad62ac9f1b3069eb Author: Jenkins Build User <tdf@pollux.tdf> Date: Thu Mar 28 15:08:33 2019 +0100 source 2b80764ec0f7c8e2c85dbca67b9cb8a2e6b2b143 source 2b80764ec0f7c8e2c85dbca67b9cb8a2e6b2b143 :040000 040000 247cc6cd4bb81a451fc03765b67a606f6930ba48 13c7cb6c2cf7bfa4e9692760972067510677d4e7 M instdir https://gerrit.libreoffice.org/plugins/gitiles/core/+/2b80764ec0f7c8e2c85dbca67b9cb8a2e6b2b143%5E!/ commit 2b80764ec0f7c8e2c85dbca67b9cb8a2e6b2b143 [log] author Noel Grandin <noel.grandin@collabora.co.uk> Wed Mar 27 14:11:27 2019 +0200 committer Noel Grandin <noel.grandin@collabora.co.uk> Thu Mar 28 13:14:11 2019 +0100 tree 4343a4c3397b9090330d89057fdf0c4440b4ac28 parent 0892ec50c2fe2ff1f26c7a4ea2fdef74e3d027d7 [diff] use unique_ptr for SwFieldTypes Change-Id: Iddfc94618e70d3ca8414d526e58746720610c552 Reviewed-on: https://gerrit.libreoffice.org/69861 Tested-by: Jenkins Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>