> I don’t know if I would consider screwing up my bibliography to be “decent”, but hey.
More or less the same how Word routinely messes up how images should be laid out in respect to the text around them. Though every office package does that to some degree. Or also how messy working with something like Apache POI is when you want to generate spreadsheets programmatically, or read them. Or how Windows keeps reverting diagnostics settings much like spyware would. Or how Linux distros have problems with sound drivers. Generally usable, good enough, but still with annoying quirks.
Then again, i'm not motivated enough to use LaTeX so aside from a bit of complaining, i guess i just have to tolerate the many packages out there and their quirks.
> You’re right about file formats - and of course it runs deeper, file formats are just a manifestation of feature sets and models. When you have a model mismatch, as LibreOffice does, then you’re sort of set up to fail, no matter how clean and open the file format is (or isn’t).
Hmm, i wouldn't do LibeOffice a dirty like that and dismiss it as some prime example of a particular bad architecture, nor would i agree that it's what my original argument was about.
I cannot comment on what would be a "good" office format example, as the internals of either look pretty bad to me, consider seeing what's inside those documents sometime. Extract the contents of a .docx and a .odt file and see the XML - Microsoft's is not quite readable, while OpenDocument's is a tad too verbose. It would probably have to be XML because of the node structure, but neither like HTML, nor what those two office packages have in store.
My original argument was closer to the following: regardless of a format being open or not, dealing with a domain such as word processing in any advanced capacity is likely to provide lots of accidental and lots of inherent complexity. Basically, any format that's more complicated than Markdown will have so many quirks and behavior that's specific to the implementation, that any other software package will be unable to reproduce it 1:1.
Just look at how many years it took for web browsers to even display CSS/HTML the same (for the most part) and they had the opportunity to work with a bunch of relatively simplistic standards, whereas office document formats feel way less developer friendly in that regard. Ergo, lacking compatibility.
What I’m trying to say is that when we say “file format” for anything non-trivial we really mean “semantic model”, the encoded representation of that isn’t so important.
There’s nothing bad about architecting software around ODF (it came first, after all) but it inevitably will lead to incompatibilities with OOXML. Likewise for the inverse. I wouldn’t call these quirks as much as an impedance mismatch.
The problems start when you have any program that tries working with a different format than the one that it was initially written for, due to all of the complexity.
More or less the same how Word routinely messes up how images should be laid out in respect to the text around them. Though every office package does that to some degree. Or also how messy working with something like Apache POI is when you want to generate spreadsheets programmatically, or read them. Or how Windows keeps reverting diagnostics settings much like spyware would. Or how Linux distros have problems with sound drivers. Generally usable, good enough, but still with annoying quirks.
Then again, i'm not motivated enough to use LaTeX so aside from a bit of complaining, i guess i just have to tolerate the many packages out there and their quirks.
> You’re right about file formats - and of course it runs deeper, file formats are just a manifestation of feature sets and models. When you have a model mismatch, as LibreOffice does, then you’re sort of set up to fail, no matter how clean and open the file format is (or isn’t).
Hmm, i wouldn't do LibeOffice a dirty like that and dismiss it as some prime example of a particular bad architecture, nor would i agree that it's what my original argument was about.
I cannot comment on what would be a "good" office format example, as the internals of either look pretty bad to me, consider seeing what's inside those documents sometime. Extract the contents of a .docx and a .odt file and see the XML - Microsoft's is not quite readable, while OpenDocument's is a tad too verbose. It would probably have to be XML because of the node structure, but neither like HTML, nor what those two office packages have in store.
My original argument was closer to the following: regardless of a format being open or not, dealing with a domain such as word processing in any advanced capacity is likely to provide lots of accidental and lots of inherent complexity. Basically, any format that's more complicated than Markdown will have so many quirks and behavior that's specific to the implementation, that any other software package will be unable to reproduce it 1:1.
Just look at how many years it took for web browsers to even display CSS/HTML the same (for the most part) and they had the opportunity to work with a bunch of relatively simplistic standards, whereas office document formats feel way less developer friendly in that regard. Ergo, lacking compatibility.