Off the Beat: Bruce Byfield's Blog

Last week, for the first time in a long while, a bug I filed saw action. I filed it over a year ago, and had almost forgot about it, which made the sudden activity around it especially welcome.

I filed Bug 98993 for LibreOffice in March 2016, as I was putting the final touches on my book Designing with LibreOfice. As I was writing, I noticed that how LibreOffice Writer handled line spacing had changed some time around release 4.2. Before then, line-spacing could be set to one-tenth of a point. After that, however, any attempt to use decimals was automatically rounded off.

If you never design documents, that change may seem trivial. After all, a point is a measurement equal to 1/72 inches, which makes a tenth of a point 1/720 inches. Surely, you might think, no one would notice such a minuscule change?

Well, actually, yes. An average font size, remember, is 10-12 points, which means that the half point when you round from 11.5 to 12.0 points is, proportionately, not so small. Use an 8 point font, as sometimes happens, and the proportion becomes even larger.

More importantly, fonts come with their own metrics for line-spacing. For a 12 point font, a common setting for line-spacing is 14.2. Rounding off, LibreOffice would display that line-spacing as 14.0. In other words, the bug meant that LibreOffice could not display many fonts the way their designers intended them to be displayed. Instead, that fifth of a point difference would make the lines in a paragraph appear cramped. Admittedly, users could change the line spacing to 15 points, but there is no guarantee that choice wouldn't make the spacing too loose.

For designers who wanted more than the default line-spacing, the situation was no better. The result of the change in line-spacing was a loss of subtlety -- and, all too often, a compromised design and considerable frustration.

An Honest Gamble
When I first became aware of the change, I thought it would soon be fixed. After all, other measurement fields in Writers used decimals, so this change would soon be corrected for the oversight it must obviously be.

No doubt I was glad of the excuse not to act. As I have written before, I am not a fan of filing bug-reports. Unless you belong to an inner circle of developers, in many projects, filing a report is an intimidating, time-consuming, and altogether thankless task. Just as the goal of insurance companies seems to be to prevent acknowledging a claim, so the goal of many projects seems to be to dismiss a report as nothing that needs action.

Still, this bug affected me directly, and nobody else had noticed it. Despite my misgivings, after stewing for a few days, I filed the bug, doing my best to write an effective report that someone else could use.

Then I waited for the splash that would indicate that the pebble I had dropped down the well had been noticed. When all the bug got was a couple of reports that it was reproducible, I shrugged and got on with other things. When activity suddenly started last week, I almost deleted unread the emails in my inbox. Why, I asked myself, should I read all the elaborate rationales for inaction?

Then I realized there was a lot of emails for a series of excuses. I clicked on the link and found that not only was the bug due to fixed in LibreOffice 5.4, but that it was scheduled to be fixed in the nightly builds for not only 5.3, but also 5.2. For some reason, a developer named Michael Stahl had decided to see the problem fixed.

I have not been in touch with Stahl, but I can imagine at least two reasons why the bug was  given attention. To start with, I imagine that changing a field to use decimals rather than just whole numbers is a trivial change.

In addition, aside from the introduction of table styles in 5.3, LibreOffice has done little over the last few years to address the needs of those who use LibreOffice the way it is designed to be used, with styles and templates rather than manual overrides. In fact, at times, I have half-believed that the general course of development was to wrench LibreOffice away from its basic design for the sake of manual formatters. Fixing this bug may have been trivial, but it gives something to designers that manual formatters may never even notice, since they rarely go so deeply into Writer as to adjust line-spacing.

Still, whatever the reasons, seeing the bug-fixed gives me unexpected pleasure. For one thing, I can now work the way I need to.

Even more importantly, though, I have finally made a small contribution to free software code. I have always maintained that writing professionally about the software takes the place of filing bug-reports, but that makes my comments easy for the bug-fixing bureaucrats to ignore -- or, if they do pay attention, I rarely hear that they have. Also, my assertion may be partly an excuse for avoiding an interaction in which I am usually at such a disadvantage that the effort feels futile.

I won't say that finally having a bug I reported fixed has changed my mind entirely about the process. However, I have studied why this one might have received attention, and I don't rule out filing more bugs -- as a free software user, I think of doing so as my duty. Considering my deep aversion to the process, that is probably the most change of which I am capable of -- but at least it is a start.