Off the Beat: Bruce Byfield's Blog

Should decisions about free software be determined by the needs of developers or users? This recurring question has become relevant again in LibreOffice 5.3, which is dropping support for Type 1 fonts.

Type 1 fonts, also known as postscript fonts, were one of the first font formats for digital typography. They have been superseded by TrueType (TTF) and OTF formats, but, unlike many earlier formats, have remained adequate for design work. As a result, many designers have continued to use them, since over the years they have a significant investment in Type 1 fonts, particularly from Adobe. At an average cost of US$35 per font family, replacing a collection made over a decade or more could cost thousands of dollars for a designer.

Now, however, adoption of the Harfbuzz font shaping engine is causing LibreOffice to drop support for Type 1 fonts. Harfbuzz does not support Type 1 fonts, and nobody has shown an interest in adding support for them. Suddenly, without any announcement that they are likely to have seen, users are opening the latest version of LibreOffice to find many of their fonts are no longer available, and their legacy documents no longer display properly. While the decision was made by LibreOfffice over four months ago, few average users are likely to follow the project closely enough to have received advanced warning. Nor did LibreOffice emphasize the change, or offer suggestions about how to react to it.

The complaints have been received without much sympathy. A bug about the change was quickly resolved by giving it a status of WONTFIX. The move is defended by noting that Microsoft Office 2013, among other pieces of software, have already dropped Type 1 support. The facts that backward compatibility has often been an advantage of free software, and that many other free software design tools such as GIMP, Krita, and Inkscape continue to support Type 1 fonts is conveniently ignored.

Instead, the blame is placed squarely on those complaining. Commenting on the bug, V. Stuart Foote suggests that any serious designer would have moved away from Type 1 fonts "some time ago if only for convenience of being compatible with workflows using TTF/OTF." Never mind that there are no compatibility problems these days between different font formats, and all three can be used in the same document.

However, it is clear that LibreOffice has no intention of accommodating the needs of users. Those complaining are described as Converting Type 1 fonts, users are told, is "trivial" with tools like FontForge -- an adjective that ignores the fact that, for those with hundreds of fonts, the time required to compensate for LibreOffice's unilateral decision is anything but trivial.

True, users can convert to other formats only as needed. However, that still means an interruption of work for the extra step. Moreover, conversion is technically illegal, if that matters to anyone.

Alternatively, users may be able to map free fonts to their proprietary Type 1 equivalents in LibreOffice's replacement tables. In some cases, the free fonts are even metrically identical to their proprietary counterparts, and can be used without any reformatting. Moreover, using only free fonts in the future may protect users from any future fiat of this sort. I have been using this tactic for several years now, and most of the time it is effective.

Yet another solution, which nobody mentions, would be to switch to OpenOffice. Switching to OpenOffice would mean sacrificing all the improvements that LibreOffice has made over the years, but the basic functionality would remain. In this case, OpenOffice's lack of developers would prove a disguised blessing, because it is unlikely to start to rip out large pieces of functionality and replace them.

However, the question remains: how dare developers put users to such inconvenience, especially without any consultation or any time to prepare? Yes, Type 1 fonts are an old format, but I suspect they remain far more common than is being cavalierly assumed.

As much as anything, what annoys me is the high-handedness of the decision and the efforts to defend it, even though I am generally unaffected myself. Users do not always understand the demands of development, but, in this case, how the decision is being carried out is as objectionable as the decision itself.