>> This isn't a *huge* problem. Word Processor documents are allowed to >> look (and layout) slightly differently between versions and between >> platforms. Of course, we should try to keep these differences at a >> minimum, but there are some other reasons why things can and will look >> differently no matter what we do: > > Here I disagree. IMO, Word Processors are *not* allowed to layout > differently, and if the user has all the fonts required, then the > document should render exactly identically, whatever version you use.
I would still disagree with you here. There are too many subtle differences that are outside of our control here, such as: * floating point operations being slightly different based on the processor used, if we use floats * font rendering programs potentially performing slightly differently * actual installed fonts differ based on platform * font licensing bits (such as apple's patents that "plague" freetype) Now, if we all used the same fonts coming from the same font rendering program with the same license in effect/not in effect, then the output should in theory be identical. In the real world and with Abi's design philosophy, we can't guarantee this. Also, I tend to lessen this requirement since a *lot* (read the overwhelming majority) of users don't care about a single pixel difference here and there in their word processing documents. It simply isn't a huge concern. >> * Different fonts on different platforms (we don't and won't embed >> fonts like PDFs) > > why "won't"? I was hopping to add something in this sense. > Of course, it's more difficult to us than to pdf, because we should > embed enough glyphs to render the text *and* to leave the user edit it. > But I see it as a desired feature (btw, word does it) I don't see this as a desired feature and only see it as causing an unending sea of headaches for us as the project progresses. Yes, PDF does do font subsetting to save size, and we wouldn't be able to do that. I'm seeing a problem when we have a 10M glyph font embedded inside of a 1 page document. Do we want a 4MB, 1 page document? Is this the best route to go? Can we be smart and say "you don't have fonts X, Y, Z. Go get them from http://myfontland.com";? What about cases where fonts are copyrighted materials? Might Abi (read me, as a US citizen and the project's maintainer) get sued because Abi now allows you to redistribute (read circumvent) copyrighted material (this is a real consideration under copyright law that we will have to be careful of, not to mention the fscking DMCA). Will this just be the responsibility of the user? I've heard the same argument applied to Napster, and they lost. You say that Word does it, which is true. Microsoft also owns a lot of the fonts, has extremely good relations (read contracts and licenses) with other font vendors, and has $40B in liquid assets and an army of lawyers to throw at problems like this. The Abi fund has $1100 and my "Lawyering for Dummies" book. If you do anything, prepare to argue your case for your proposed solution, because there will be arguments for and against it, and I think that most arguments for all sides will have a 1+ kernels of truth to them. >> I personally don't care if you use floats or integers, but please be >> consistent. I think that's more important than anything else. > > I was a bit more pro-float, but seeing Jody post (ie. this issue may be > more important for tables than for simple text) puts me in a middle > zone. > > Also the fact that TeX picked integer calculations, getting a > formidable > layout stability over the years has a certain weight in my mind. Integers could be good. Then again, we're talking about word processing documents here with giant columns, lots of lines, and lots of space in between them. We're also dealing with users who tend to be much less picky if a word gets wrapped differently on Win32 than on Unix, if they move documents between platforms at all. Jody's spreadsheet problems are at least due to the fact that he's dealing with much less real-estate. His cells are 1 inch by 1/2 inch at most. The margin for error is much higher there, and you absolutely must get the cell's contents right every time, no matter what. No matter what choice you make, I don't think your decision (or our results) will be unprofessional. Dom
