OOops: Why OpenOffice.org* Isn’t Nearly Good Enough

* Et tu, LibreOffice

I have lived through a fair number of Year of the Linux Desktops, and consequently had the pleasure of using tolerating OpenOffice.org for years. In the light of recent news of the German Foreign Office migrating back to Windows XP from Linux, I thought of writing about one aspect this ‘interoperability’ that the German FO is crying about.

I never felt OpenOffice.org to be lacking in any respect compared to Microsoft Office 2003 when I was in high school (back in the day, Office 2007 wasn’t out yet). OOo satisfied my requirements without costing me a dime. In hindsight, that was only because on most occasions I was printing out hard copies of documents for submission, and didn’t often have to bother about whether a document that was displaying as I wanted on my PC would display the same way on another.

Now, in university, interoperability is a major headache for me. And yes, by ‘interoperability’ I mean the very narrow of definition of ‘whether document files work with Microsoft Office’. When working on project reports or some such, I often have to work on documents that have images, charts, tables formulas et al, i.e., fairly ‘complex’ formatting in as to how any element on the page is placed. More often than I like it, I find the formatting screwed up. Even for something as simple as taking a document to the library for printing, it has become second nature for me to export my ODF documents to PDF so that I have ‘assured’ formatting on library (Windows) machines.

Let me illustrate my points with a few examples. I am the studio director at the student TV station in my (exchange) university and I need to edit studio scripts every week that is sent to me by the production assistant, edited on her Office 2010 install. This is how it looks when opened in Microsoft Office 2010 Starter Edition on my netbook.

When I open the same file in OpenOffice.org 3.3.0, out of that 15 page document only TWO pages are displayed. I am not kidding. It actually shows up as ‘Page 1 / 2’. I can understand this is an Office 2007/2010 .docx format document and support for this “isn’t as good yet“, but jeez, showing 2 pages out of 15 is a major screw-up, isn’t it! Let’s see what out of the pages it can read does OOo show.

On surface, it seems that OOo got most of the page displayed correctly, but a closer look will reveal page elements that are completely missing. The ‘SOVT’ icon is gone, so is the box with ‘S/I’, as well as adornments such as boxes around ‘ZOOM 5sec’. If I had to solely rely on OOo for my document needs, I wouldn’t have known these elements are missing. For me, this is a major problem as those are crucial directives that I need when directing the studio shoot.

Even if I concede that this problem may be because OpenXML file support isn’t as ‘complete’ yet in OOo, that doesn’t explain why it utterly messes up page display with older .doc files, like this example below.

Content from two consecutive pages was stacked on top of each other on one page along with previous edits even though show changes was turned off, rendering this file useless. Note that this file was created in Office 2007 and exported as .doc.

Okay, so maybe the whole problem is with Office 2007/2010 creating corrupted files which it itself can figure out, but not other software. Maybe the files that I was trying to open was ‘too complex’ (debatable). How about a simple document – a table with two columns with text in rows – created in a word processor not developed by Microsoft – say, Mac OS X’s bundled TextEdit – and exported as a .doc file. Like this file below.

This TextEdit-created file displays correctly in Microsoft Office 2007/2010 too. And here’s how the same file displays in OpenOffice 3.3.0.

It isn’t as if these are two isolated incidents where OpenOffice.org has failed me in being passably reliable, even for freeware software. I have lost track of the number of times when my documents/presentations have been messed up by OOo. Sometimes, opening a .docx file created and saved using OpenOffice.org fails to render correctly on the same system with the same install of OOo. Figure that. And while Microsoft may have a conflict of interest, Doug Mahugh makes a seemingly well-reasoned argument how certain aspects of ODF file rendering is broken among variants of OOo itself like IBM’s Lotus Symphony.

Throughout this blog post, I have mentioned OpenOffice.org as the errant document suite, but the new Document Foundation’s ‘LibreOffice’ fork isn’t any better. A few weeks ago, I was working on a presentation that I needed to send for a job interview. I created it in OpenOffice 3.3.0 one day (saved as ODP) and picked up work the next day on a new install of LibreOffice 3.3.1. (The Document Foundation advises users to remove OOo before doing a LibreOffice install.) I saved a PPT and a PDF version of the file in addition to the native ODP file (my usual routine) after I was finished. I opened the file Microsoft PowerPoint 2010 Viewer to check before sending it off, and got an error message telling me that the file was corrupted. Perplexed, I tried opening the file again in LibreOffice only to find that LibreOffice couldn’t open it. What’s more, when I tried to open the ODP file for the same, LibreOffice couldn’t read that either! The only way I got out of this pickle was because I had an old version of the ODP file saved in my Dropbox account. (Thanks, Dropbox, for maintaining copies of deleted files forever!)

A quick browse through the bug tracker for both OpenOffice.org and LibreOffice shows a sizeable number of bugs that affect a (significant?) subset of users with document display. In most cases it seems there’s a problem with a particular file or that it cannot be consistently reproduced across a number of systems. But it’s precisely this uncertainty that I dislike – and I think I speak for a lot of OOo/LO users when I say this. I hate having to always export a PDF version so that I can email to friends (for one, they can’t edit it) because I can never be sure enough that it’ll display exactly the same if I send a .doc/.odt file. I hate constantly having to dual-boot into Windows just to edit a document, like I often have to do.

Going back to the German Foreign Office’s decision to switch to Windows, I can empathise how poor word processing / spreadsheet support can be a show-stopper for a major government division since OpenOffice.org is the only document editing suite that comes anywhere close to functional on Linux. (Don’t kid yourself. Abiword and Gnumeric don’t count.) And yet, there is indignation among the FOSS community calling the reason for switching ‘implausible’ and much vitriol thrown in the form of users-are-stupid / Windows-is-bloatware ad hominem attacks.

Part of the problem I think has its roots in hacker-developers writing software for the masses. Anecdotal evidence suggests that hacker-developers often aren’t heavy users of document suites, so maybe they don’t really feel the pinch in a way general users do. There are usability bugs such as this (in OOo) that have been around for seven years with no resolution. They maybe be perfectly happy typing out documents in LaTeX on vim so their definition of what’s a show-stopper differs very much from a lay user’s definition of show-stopper.

LibreOffice has a well-defined release criteria to define what type of bugs can be considered to be ‘release blockers’, and while in their books document rendering compatibility is a not a ‘major’ issue, for a lay user it is. I understand that releases cannot be delayed forever and if releases are held off forever till everything is polished to perfection then nothing will ever come out. The rationale is that what’s crucial to one subset of users may not as be important for a second subset of users. My feeling is that Joel Spolsky gets it bang on target when he says:

A lot of software developers are seduced by the old “80/20” rule. It seems to make a lot of sense: 80% of the people use 20% of the features. So you convince yourself that you only need to implement 20% of the features, and you can still sell 80% as many copies. Unfortunately, it’s never the same 20%. Everybody uses a different set of features.

I don’t know whether I belong to a minority that requires solid table rendering support while most others get along fine without it, but it could be that the bugs or missing features turn people off from using the software in the first place. Because I know that OOo/LO handles formatting incorrectly and sometimes straight up eats data, I keep a dual boot install of Windows 7 just so that I can edit documents in Office 2010 Starter. I put up with it, but how many other users would you expect to?

Therein lies the core problem for OpenOffice.org and its soon-to-be-prevalent fork LibreOffice: as long as there’s the slightest hint of doubt on how well it interoperates with the dominant Office ecosystem, they will never be the preferred choice for business/enterprise or even personal-use scenarios. It will always remain a “…but yeah we can save some money as long as we can live with some unreadable documents”. Actually, that goes for competitors like Google Docs too that have similar compatibility issues. I would never consider using Google’s Chrome OS as all that Google Docs can handle are simple formatted documents. Setting expectations right should also be taken into consideration; if the LibreOffice install that corrupted my presentation was a beta or an RC, then I’d have been more cautious, expected certain things to not work. But it wasn’t – 3.3.1 was a final release. What’s the rush to release a final version instead of longer beta testing periods?

The FOSS community on the whole needs to listen to voices who are coming forward to say, “Look, this doesn’t work for us.” like on this Slashdot thread. Admitting the fact that there are areas in which features are lacking is the first step towards making progress rather than making remarks that can be considered callous or snide by new adopters. What OpenOffice.org/LibreOffice need to do is deal with even minor annoyances like Canonical did with its 100 Papercuts project. (I applaud Canonical for caring about users and taking steps such as this to improve usability overall.) It’s encouraging that LibreOffice is already taking a step in this direction.

Still, it’s early days in LibreOffice’s existence. The reason the project forked from OpenOffice.org was a culmination of a long festering resentment of how the development was proceeding and there were often political reasons why changes were not contributed back to OOo (Go-oo, etc). I hope that under new leadership and new supporters such as Canonical and Novell, LibreOffice develops into a much more mature offering.

Until then, I can only pray that I can read/write files without a hitch.