Download it now!
Bug 127589 - Filesave XLSX: Missing data in two graphs that were shown before
Summary: Filesave XLSX: Missing data in two graphs that were shown before
Status: NEW
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Calc (show other bugs)
Version:
(earliest affected)
5.3.0.3 release
Hardware: All All
: medium normal
Assignee: Not Assigned
URL:
Whiteboard:
Keywords: bibisected, bisected, dataLoss, regression
Depends on:
Blocks: OOXML-Chart XLSX
  Show dependency treegraph
 
Reported: 2019-09-17 08:05 UTC by Timur
Modified: 2020-10-08 08:51 UTC (History)
5 users (show)

See Also:
Crash report or crash signature:


Attachments
Example compared in MSO and different LO after save and open round-trip (173.06 KB, image/jpeg)
2019-09-17 08:05 UTC, Timur
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Timur 2019-09-17 08:05:08 UTC
Created attachment 154212 [details]
Example compared in MSO and different LO after save and open round-trip

If we *open* XLSX attachment 102146 [details] from Bug 80816 it's mostly fine, except some format issues, most notably title text format.
MSO opens as PDF attachment 102601 [details]. 
If we *save* as XLSX in LO, there used to be more differences, as shown in PDF attachment 102604 [details].
But now in LO 6.4+, data in two graphs is completely missing on filesave and that's this bug. 
I'd call this a regression.
Comment 1 Timur 2019-09-17 08:08:20 UTC
LO 5.2 used to show data, 5.3 no.
Comment 2 Gabor Kelemen 2019-09-17 08:14:22 UTC
(In reply to Timur from comment #0)
> If we *open* XLSX attachment 102146 [details] from Bug 80816 it's mostly
> fine, except some format issues, most notably title text format.

FYI that is known as bug #78027
Comment 3 Xisco Faulí 2019-09-24 11:43:33 UTC
Regression introduced by:

https://cgit.freedesktop.org/libreoffice/core/commit/?id=02af87fdd76bc94fb51aeb160c74d6f719c42c63

author	Eike Rathke <erack@redhat.com>	2016-10-27 12:52:01 +0200
committer	Eike Rathke <erack@redhat.com>	2016-10-27 12:52:34 +0200
commit	02af87fdd76bc94fb51aeb160c74d6f719c42c63 (patch)
tree	46a2371ec6fa1daf8d9873c3621a9c258ea2ea08
parent	91114f68e0a81d4f2a5354bc6057f62c22c780b4 (diff)
Resolves: tdf#103531 OOXML: write external file ID within quoted sheet names

Bisected with: bibisect-linux-64-5.3

Adding Cc: to Eike Rathke
Comment 4 Xisco Faulí 2020-02-12 16:07:28 UTC
Still reproducible in

Version: 7.0.0.0.alpha0+
Build ID: 718f540fb63af27c1336f89213444e9af753b8a9
CPU threads: 4; OS: Linux 4.19; UI render: default; VCL: gtk3; 
Locale: en-US (en_US.UTF-8); UI-Language: en-US
Calc: threaded
Comment 5 Eike Rathke 2020-02-12 18:58:59 UTC
Taking.
Comment 6 Eike Rathke 2020-02-13 13:22:24 UTC
@Xisco: The cached external data here is not written to OOXML, hence when reloaded all data in the chart is blank; that mentioned commit is not the culprit, test-reverting the change it still fails.
Could I have an actually matching bisect please?

I have an idea what's missing, but maybe there's an indicator why and why it worked before, if it really did.. unfortunately my earliest local build is 5-3-7 so I can't check, compiling that old code with a new compiler doesn't work without modifications.

It may of course be that things worked differently in the past and the commit somehow made it fail and then afterwards things were changed so that it now doesn't matter and fails anyway, so bisecting correctly misleads.
Comment 7 Eike Rathke 2020-02-16 19:09:04 UTC
The same in Excel 97 .xls binary format. The external data is only referenced in the chart data ranges, not in sheet formulas, hence not added to the XclExpSupbookBuffer. I doubt this ever worked, but if so would be interested in the real bisect.
Comment 8 Eike Rathke 2020-02-16 23:29:54 UTC
Unassigning for the time being.
Comment 9 Xisco Faulí 2020-02-18 11:27:51 UTC
Hi Eike,
I just bisected it and it points me to the same commit ( either using the UI or command line... --convert-to ). I double checked using git checkout HEAD~1. OTOH, the problem with xls files can be reproduced with previous versions
Comment 10 Xisco Faulí 2020-02-18 11:30:54 UTC Comment hidden (obsolete)
Comment 11 Xisco Faulí 2020-02-18 11:38:04 UTC
(In reply to Xisco Faulí from comment #10)
> @Buovjaga, @Raal, would it be possible to bisect this issue on windows to
> see if it points to the same commit ? (
> 02af87fdd76bc94fb51aeb160c74d6f719c42c63 in win-5-3 repo )

Actually I had that repo on win so I bisected it too. Both linux and windows repositories point me to the same commit
Comment 12 Timur 2020-02-18 12:18:42 UTC Comment hidden (obsolete)
Comment 13 Aron Budea 2020-02-29 08:29:08 UTC
(In reply to Timur from comment #12)
> Sorry for the basic question, but does bibisect commit has multiple code
> commits or just that one?
Just this one.

Also, the backported commit (bug 103531 comment 4) can be found in bibisect-linux-64-5.2, and is the cause of the same regression in that repo.