Advogato blog for caolan

dead code list

Wed, 7 May 2008 07:09:02 GMT

Up to date lists of the currently unreferenced symbols in OOo are available here. Currently 2823 methods, though a lot of these should go away when the submitted patches filter through.

OOo gio integration

Thu, 1 May 2008 11:16:53 GMT

So I finished off Jan’s initial gio OpenOffice.org integration work (i.e. issue 88090 as workspace ucpgio1)

The gio api is a far better fit to the OOo worldview than gnome-vfs is and was pretty straightforward to implement modulo my own bafflement as to why I had an error claiming that mounting share named “X” failed as not existing which incredibly had the root cause that the share named “X” was really called “Y”.

GoOOoCon08

Mon, 14 Apr 2008 10:11:50 GMT

The GoOOoCon08 Presentation that I gave.

ms binary docs

Tue, 19 Feb 2008 15:05:52 GMT

Since the ms binary docs documentation release I’m gotten a veritable blizzard of mails about it.

The move is welcome of course, efforts to improve interoperability are always appreciated. So thanks.

But the original ‘97 formats were released on MSDN around 97/98 and were available on the MSDN website for some months around then, so this isn’t as totally new as people seem to think. MS has done this once before. (I wrote ivt2html a stack of years ago just to read the .ivt format of that era). Though clearly the documentation for 2007 will include the changes since 97, but the changes to the format are compatible ones and over the years OOo has already figured out pretty much all of the relevant additions. So the release of the formats is likely to help other projects that want to start from scratch much more than it helps a project like OOo that has pretty much already figured out the majority of what I assume (I don’t do much .doc import/export anymore) is in the documentation.

The remaining major issues for compatibility IMO for .doc/.odt at least break down into 3 categories

Misunderstanding of the format, that’s the smallest issue by far, maybe some table in table glitches fall into this category
Disjoint feature sets, e.g. there are some constructs in writer that don’t exist in word which are problematic to export to .doc, i.e. all of the writer page styles possibilities are not expressible in the word section system, and vice-versa there are word features that don’t have mapping in writer, i.e. highlighting as a separate setting than text foreground color, though that’s obviously easier to fix
weirdness, i.e. the layout algorithm rules to determine what to do when faced with layout constraints that cannot be met, e.g. roughly a circular dependency where e.g. a graphic anchored to a paragraph affects its own layout in some nasty feedback way, especially in multi-column documents. As as example of an oddity that’s taken care of by the .doc import/export filters the top left corner of a graphic in writer is position x,y of its properties and the border (if any) is drawn inside that extent while in word the border is drawn outside it, i.e. the top left corner of a bordered graphic’s exent in word is x-borderwidth,y-borderheight, unless it is one of a small class of banded borders where only the first stripe or two is positioned outside the unbordered graphics extent, and the other stripes are inside it. i.e. when importing or exporting it you have to know what type of border is being applied and fudge the figures to get the same size and position of the total entity to get the same positioning as in the other application.

OOo ia64 port

Sun, 17 Feb 2008 12:11:32 GMT

I’ve completed a ia64 linux uno bridge. Patch here and corresponding demo OOo debs.

24 Jan 2008

Thu, 24 Jan 2008 10:07:21 GMT

Here’s a zoomed in view of the gedit File label which is rendered with cairo. . And here’s one of the same label as rendered by OOo currently.

After adding my experimental cairo text render patch, here’s the OOo one again. . Pretty neat.

Cairo makes it utterly trivial to support text rotation, scaling and artificial oblique. All pretty neat. On the other hand currently artificial emboldening is painful as you can only get it through the cairo_ft_font_face_create_for_pattern api and not my preferred cairo_ft_font_face_create_for_ft_face, so the code to do that is disabled and we fall back to the current rendering mechanism for emboldened fons. You can turn it on in today’s rawhide OOo with export SAL_ENABLE_CAIROTEXT=1. I haven’t done any extensive testing to see if all is ok with CJK text, but I don’t really expect any problems.

pixmap leaking

Mon, 21 Jan 2008 15:05:47 GMT

So I fixed a little Pixmap leak in OOo and afterwards wanted to see what the the remaining pixmaps might be. Stripping down this code to gave me a quick and dirty Pixmap browser that takes pixmap ids from stdin and then I stuffed a LD_PRELOAD lib in there to intercept OOo’s XCreatePixmap and XFreePixmap and fed the list of unfreed pixmaps to the browser and ta-da intsta-handy debugging tool. pixmaptracker and browser.

english locales

Wed, 9 Jan 2008 11:13:13 GMT

For some reason (crap translations ?, just prefer using English ?), it’s common for some non-english speakers to run their desktop with an en_US locale but then to have difficulties that stuff like openoffice.org, etc. defaults to letter size paper and inches for display units. So my modest suggestion is that if you want to use an English locale then a non en_US locale like e.g. en_GB will default to the way more common A4 paper format and also default to cm for units. Even better for most europeans, using en_IE will also give you A4, cm and additionally the default currency as Euro.

hp ski simulator

Mon, 3 Dec 2007 16:04:10 GMT

Tried to get the hp ski ia64 simulator running recently, and fell into some newbie problems trying to get a 2.6 kernel running at the right speed with a useful amount of memory on it, mostly because I was looking at the wrong bit to try and find what limited the memory and controlled the clock timing, i.e.

The bootloader is what restricts the available memory, not the simulator itself. Patch to increase from 130Megs to 1Gigs
The clock runs way too fast, again the simulator doesn’t control that, the bootloader does. Patch to reduce speed dramatically so make doesn’t get woefully confused
A prebuilt 2.6.18 kernel, and bootloader from this config

Seagate Free Agent Pro

Thu, 29 Nov 2007 21:04:37 GMT

Amazingly,