GNOME Bugzilla – Bug 465545
Crashes for 64 bits non-GNOME build
Last modified: 2007-08-13 17:50:17 UTC
I am using gnumeric (currently 1.7.11 on linux - goffice 0.4.2) on my laptop x86. It is working pretty well for a test version with some bugs here and there but easy to turn around. Unfortunateley, I can't use it on my 64 bits desktop. Here is the main symptoms: 1. I need to delete .gnumericrc file before each run! If I don't, it used to crash. For 1.7.11 it starts but with most cells displaying "#Name" (or something - I am using the pt version). I Also include below an extract of the console output when I call gnumeric without deleting .gnumericrc. It is a very huge file to put all of it here. 2. I have a big "book" that depends on a sheet which I fill using copy/paste special value from another sheet in another book. This operation sometimes crashes the destination "book". I will try to get more info on this (because of its unstability on x86_64, I don't use it much on this platform). None of these problems occur for x86 32 bits. Regards and thanks for what is already a good program. _____________________________________________________ ** (gnumeric:10482): WARNING **: Using default value '1' Reading file:///home/p/EM_CURSO/EstInvest/t/AllData.gnumeric ** (gnumeric:10482): WARNING **: unknown type 'GogHistogramPlot' (gnumeric:10482): GLib-GObject-WARNING **: cannot create instance of abstract (non-instantiatable) type `GogPlot' ** (gnumeric:10482): CRITICAL **: gog_object_add_by_name: assertion `IS_GOG_OBJECT (parent)' failed Unexpected element 'line' in state : Workbook -> Sheets -> Sheet -> Objects -> SheetObjectGraph -> GogObject -> GogObject -> GogObject -> GogObject -> property Unexpected element 'fill' in state : Workbook -> Sheets -> Sheet -> Objects -> SheetObjectGraph -> GogObject -> GogObject -> GogObject -> GogObject -> property ** (gnumeric:10482): WARNING **: unknown type 'GogHistogramPlot' (gnumeric:10482): GLib-GObject-WARNING **: cannot create instance of abstract (non-instantiatable) type `GogPlot' ** (gnumeric:10482): CRITICAL **: gog_object_add_by_name: assertion `IS_GOG_OBJECT (parent)' failed Unexpected element 'line' in state : Workbook -> Sheets -> Sheet -> Objects -> SheetObjectGraph -> GogObject -> GogObject -> GogObject -> GogObject -> property Unexpected element 'fill' in state : Workbook -> Sheets -> Sheet -> Objects -> SheetObjectGraph -> GogObject -> GogObject -> GogObject -> GogObject -> property ** (gnumeric:10482): WARNING **: unknown type 'GogXYPlot' (gnumeric:10482): GLib-GObject-WARNING **: cannot create instance of abstract (non-instantiatable) type `GogPlot' ** (gnumeric:10482): CRITICAL **: gog_object_add_by_name: assertion `IS_GOG_OBJECT (parent)' failed Unexpected element 'line' in state : Workbook -> Sheets -> Sheet -> Objects -> SheetObjectGraph -> GogObject -> GogObject -> GogObject -> GogObject -> property Unexpected element 'marker' in state : Workbook -> Sheets -> Sheet -> Objects -> SheetObjectGraph -> GogObject -> GogObject -> GogObject -> GogObject -> property ........................................... cut here ** (gnumeric:10482): WARNING **: [AxisBaseView::render] not implemented for this axis set (-1) ** (gnumeric:10482): WARNING **: [AxisBaseView::render] not implemented for this axis set (-1) ** (gnumeric:10482): WARNING **: [AxisBaseView::render] not implemented for this axis set (-1) ** (gnumeric:10482): WARNING **: [AxisBaseView::padding_request] not implemented for this axis set (-1) ** (gnumeric:10482): WARNING **: [AxisBaseView::padding_request] not implemented for this axis set (-1) ** (gnumeric:10482): WARNING **: [AxisBaseView::render] not implemented for this axis set (-1) ** (gnumeric:10482): WARNING **: [AxisBaseView::render] not implemented for this axis set (-1) .............................................. EOF
Andreas and I both run on 64-bit machines. So things are regularly tested there and something else is up. It appears like you are not getting plugins loaded. Could you try doing a "make clean", "make 2>&1 | tee logfile", and "make install"? Please look for warnings in logfile.
(In reply to comment #1) > Andreas and I both run on 64-bit machines. So things are regularly tested > there and something else is up. > > It appears like you are not getting plugins loaded. None of these messages appears if I delete ~/.gnumericrc file before invoking gnumeric. In this case gnumeric seems to run fine for most things except for the copy/past problem I reported that sometimes occur (only on 64 bits). BTW, I have just done the following steps: Delete ~/.gnumericrc gnumeric Enter A1: 1, A2: 2, A3: 3, B1: =min(A1:A2) Save as TP.gnumeric exit gnumeric gnumeric TP.gnumeric Cell B1 displays #NOME? (NOME is the portuguese word for NAME) exit gnumeric Delete ~/.gnumericrc gnumeric TP.gnumeric Cell B1 displays 1 (correct) In *all* this example only the following message is displayed when invoking gnumeric: ** (gnumeric:8215): WARNING **: Using default value '1' Reading file:///home/psergio/tmp/TP.gnumeric > > Could you try doing a "make clean", "make 2>&1 | tee logfile", and > "make install"? I am using gentoo (up to date), kde environment, no doc, and used an ebuild I found in gentoo bugzilla, the place where new ebuilds are proposed. Exactly the same environment and ebuild is/was used for my 32 bits system where this problem never occured. I'll see if I can build it manually or get the compilation output from the ebuild. Thanks.
Created attachment 93468 [details] ebuild output gnumeric 1.7.11
I have run the ebuild and got its output I am attaching here. The link to the "bug report where the ebuild is": http://bugs.gentoo.org/show_bug.cgi?id=156984 If you need any further info pls. let me know.
hmm, I have a gentoo amd64 box at home, so I might test the ebuild, but I'll not go home before september now.
The ebuild is not enough, which use flags do you use ( may be -gnome?)
and your .gnumericrc might be insightful too.
(In reply to comment #5) > hmm, I have a gentoo amd64 box at home, so I might test the ebuild, but I'll > not go home before september now. > For me it's fine. I would prefer to use my 64 bits desktop in a remote session mainly because it has a lot more performance (+mem and ++HD speed), but I can leave with my laptop. In the meanwhile may be the situation gets fixed. Before 1.7.11 gnumeric never get started without deleting .gnumericrc first.
Created attachment 93491 [details] USE flags
(In reply to comment #6) > The ebuild is not enough, which use flags do you use ( may be -gnome?) > I created an attachment with all my USE flags. BTW, my 32 bit system has exactly the same USE flags and, except for 64 bits specific SW and versions, it has the same environment.
Created attachment 93494 [details] .gnumericrc
This problem has been fixed in the development version. The fix will be available in the next major software release. Thank you for your bug report. (That was a problem in the non-gnome build. This build is not tested very often.) Compiler warnings made this obvious! This would have been easier if you had attached the compilation log.