Apache OpenOffice (AOO) Bugzilla – Issue 6522
Open plain text file: Error loading converter
Last modified: 2013-08-07 14:43:39 UTC
When I try to open a plain text file in OOo I get the following error: Error loading document file:///<name> Read-Error. Error loading converter. I am also unable to paste plain text from the clipboard. Cntrl-v does not work or paste greyed out normally. Sometimes a clipboard format not supported error is raised. OOo 1.0.1 installed under SuSE 7.3. David.
This is two issues within one description. That's impossible to handle. Being more precise than "sometimes" and attaching a document showing the fi5st one would ease evaluation too. Please choose a "survivor" and write a new issue for the other one in case you can reproduce it in the latest build. Thx.
Created attachment 4190 [details] Plain text file with just two lines of text
OK I'll file a separate report for the plain text paste problem. The problem with opening a plain text file is still present in OpenOffice.org 643C for Linux. I've posted a sample text file that gives the error exactly as reported in the original report. Tested using OOs 643C for Linux running under SuSE 8.0, KDE 3.04. David.
Reassigned to Ulf.
...investigating. Although I didn't get the error message on loading the bugdoc, I had some crashes on loading plain text files in the default encoding (western iso8859-1) but was never able to reproduce this. Ever since I started OOo once in a debugger no more crash happens. I noticed this behaviour with several versions; OOo1.0.1 until current (master) inhouse versions. us->cmc: did you also encounter suspicious behaviour on text files?
Hmm, no not with "normal" text files. I did see totally bizarre behaviour with asian text files under solaris/linux once and that got tracked down to a font based glitch of somekind outside the filter. Can't see that being the problem here
Hi, I've just realised that this report is the same as some that are in the Debian bug tracking system. I had not reported it here because I had always thought that it was Debian-specific, but David says that he is using SuSE 7.3 so the problem must be more widespread. The bugreport logs can be found here: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=173481 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=182452 Here is a summary of the information in the logs: Date: Mon, 13 Jan 2003 10:33:16 -0500 From: Adam Kessel <adam@bostoncoop.net> I've now figured out how to fix it, but not why it was happening. Previously I had my locale set to "C" and no locales generated, by reconfiguring locales (dpkg-reconfigure locales) and generating en_us and selecting that as my default locale, the problem goes away. Now I can no longer reproduce the bug, even if I switch the locale back from en_us to C and uncheck en_us to be generated by locale. But I'm certain that this caused the bug to disappear. I've answered your questions below. On Mon, Jan 13, 2003 at 03:05:03PM +0100, Chris Halls wrote: > On Mon, Jan 13, 2003 at 07:51:22AM -0500, Adam Kessel wrote: > > I get this error (text error filter) with all text files. I am thus > > unable to open any text files in OpenOffice. > We are unable to reproduce this bug at all, so we need detailed information > to help us reproduce the error. Please will you tell us the following > information: > 1. What exactly is the error? Word for word. What did you do to reproduce > it? There are several ways to open a document; does the error occur > whichever way you open the document? > - Click on an openoffice icon in the menu, File -> Open > - On the command line, type 'openoffice <filename>' Error occurs both from the command line and from the menu. The same error appears in both places: Error loading document file:///home/adam/file.txt:=20 Read-Error. Error loading converter. > 2. When you open the document from the command line, what messages do you see? ~>openoffice file.txt=20 Gnome session manager detected - session management disabled I18N: Operating system doesn't support locale "" (FWIW, my locale was set to "C"). > 3. If you create a new user and start openoffice as that user, do you still get the problem? Yes, same error with new user. > 4. If you reinstall the openoffice.org and openoffice.org-bin packages, do you still get the error. Are there any error messages at installation time? Since I "fixed" the problem as above, I can't test this. > 5. Do you have another machine available? If so, is the error reproduceable there? Not reproducible on other machines (presumably because their locale is set). ---------------- To: halls@debian.org Subject: Re: Bug#182452: openoffice.org: can't import plain text From: Rich Walker <rw@shadow.org.uk> Date: 27 Feb 2003 02:48:49 +0000 Message-ID: <m3lm02o2um.fsf@tigger.shadow.local> > Hi, thanks for reporting; this bug is already reported and the previous submitter was able to get OOo working like this: > -------------------- > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=173481 > > I've now figured out how to fix it, but not why it was happening. > > Previously I had my locale set to "C" and no locales generated, by > reconfiguring locales (dpkg-reconfigure locales) and generating en_us > and selecting that as my default locale, the problem goes away. > > Now I can no longer reproduce the bug, even if I switch the locale back > from en_us to C and uncheck en_us to be generated by locale. But I'm > certain that this caused the bug to disappear. > -------------------- > > Does the problem go away for you too if you follow this procedure? Okay, after doing the "locale is en_GB.." thing earlier I've now logged out and logged back in (under icewm rather than kde...) and the problem has gone away. Hmm. Maybe the openoffice I18 message about the locale not being supported could be made a little more verbose - like, "This may cause weird problems. Fix by ..." cheers, Rich. ---------------- This fixed the problem. It isn't in the openoffice.org FAQ, or the debian docs, as far as I can tell. However, OOo was generating a locale error - maybe the locale error should tell us this. cheers, Rich. -----------------
Running linux-2.4.20 and OOo-1.0.2/OOo-1.1beta A little more investigation shows that I can open *SOME* plain text documents with OOo and problem documents may be due to line termination issues. For example, this document: [john@starfleet palm]$ file plrabn12.txt plrabn12.txt: ASCII English text, with CRLF line terminators opens fine in OOo, but this one: [john@starfleet palm]$ file candide.txt candide.txt: ASCII English text, with very long lines does not open. Nor does this one: [john@starfleet palm]$ file rgain10.txt rgain10.txt: ASCII English text Now, if I crank the file rgain10.txt through "unix2dos" to add CRLFs to the file, it works: [john@starfleet palm]$ unix2dos rgain10.txt unix2dos: converting file rgain10.txt to DOS format ... [john@starfleet palm]$ file rgain10.txt rgain10.txt: ASCII English text, with CRLF, CR line terminators And if I run candide.txt through unix2dos, it also now works with OOo: [john@starfleet palm]$ unix2dos candide.txt unix2dos: converting file candide.txt to DOS format ... [john@starfleet palm]$ file candide.txt candide.txt: ASCII English text, with very long lines, with CRLF line terminators Is this something that can be fixed in the converter?
US->JDThompson: unfortunately you didn't attach one of your bugdocs. Pls. do so as I still can't reproduce. Thx.
Received bugdoc by private mail: bugdoc opens fine in OOo 1.1 Beta1 and OOo 1.0.2. Locales set to 'en_US' or 'C'. But when setting the locale to 'foo': export LC_ALL=foo; export LANG=foo the described error message pops up. Thus I guess the text filter needs a valid locale to be set to work correct. US->CMC: is it feasible for the filter to detect an invalid locale and perform a fallback to 'C' or alternatively provide the "ASCII Filter Options" dialog?
Ive had this problem with Lycoris (based originally on Caldera's Workstation 3.1.1). Changing the locale (Tools > Options > Language Settings > Language) to US_en (as opposed to "default") had no effect. The problem remains. Running unix2dos on a .txt file opens the ASCII Filter setup where I can choose the encoding (Western European) and the font (both must be chosen to prevent an error). The file then opens, but only as "Read Only." Nothing further seems to help.
cmc->mmaher: As discussed.
MM: Started looking at this...
MM: Changed parasc.cxx in the ascii filter so that if the locale uses an unregonised CharSet it defaults to ASCII
This is listed as "Fixed" -- will it be fixed in 1.0.3 or should I wait for another release?
MM->RBE: This fix will be available in OO 1.1 beta ( I think early May ).
verified fix in jollyfilterteam04.
verified in jollyfilterteam04.
Closed. Fix available in OO 1.1 Beta2.