LibreOffice 4.4 To Have A New OpenGL Back-End

[wikinevick]

The sole use of an office suite isnt just "How well does this render another company's files" if you want that then just get a viewer. A full office suite also has to handle creating files-- which if you hadn't just revealed your ignorance by showing you hadn't read the attached blog post-- you'd know that this effects the way you embed and handle images into documents and most importantly is one of the steps necessary to get LibreOffice working on Wayland.

You are completely missing the point on why the German cities are moving back to MS-Office (and Windows). The fact is that humans are social by nature and you can be expected to open "another company's files", edit them, and send them back. And they must absolutely render fine on that other company's" leading commercial editor.

You can sometimes get away with .pdf though, but most of the times, just having an old version of MS-Office works better that LibreOffice. And of course, the one thing LibreOffice needs to catch more users is change the name.

[quikee]

Indeed. The differences between LibreOffice and OpenOffice are rather minimal and neither of them has a killer feature. LibreOffice seems to be taking all the features from OpenOffice but is somewhat more unstable, perhaps a sign of the lack of coordination between both projects

Minimal things like writing OOXML and stuff like that.

[quikee]

You are completely missing the point on why the German cities are moving back to MS-Office (and Windows). The fact is that humans are social by nature and you can be expected to open "another company's files", edit them, and send them back. And they must absolutely render fine on that other company's" leading commercial editor.

You can sometimes get away with .pdf though, but most of the times, just having an old version of MS-Office works better that LibreOffice.

Sure, MS Office compatibility is very important (and is being worked on practically constantly - there is something in every LO release) but this doesn't mean that other things aren't important and should not be worked on. Actually this is a huge step forward as VCL backend code (and things like timers) needed some love for a long time now and it opens doors to new posibilities in the future (I think Impress could benefit the most from this changes).

And of course, the one thing LibreOffice needs to catch more users is change the name.

Because people will start take you more seriously when you change your name..

[Remdul]

Afaik moving to OpenGL is much more difficult than the author believes.
1. Afaik what you do on the CPU is perfect quality, while the GPUs are more sloppy about quality and following the GL specs, especially the older ones.

GPUs are no longer as sloppy as they used to be. All modern GL HW implementations by now have things like exact rasterization of lines, texture filtering (finally) worked out, and is consistent. Even Intel drivers. There is still some shader stuff that becomes undefined, but only in really complex situations where you're really pushing the boundaries of the API/GLSL/hardware capabilities. Not something used in OO/LO, or typical user interfaces anyway.
Then there is GLES, which is far more specific on the little details, and there are conformance tests. There will always be drivers that mess up, but I think it is now fair to move on with it. We all laughed at Intel, but they've caught up that it is reliable enough. And anyone left with crappy GL implementations shouldn't be taken serious and hold back porting UIs to GL.

2. OpenGL speed is a valid point for medium or big tasks (like image scaling), but smaller tasks like drawing a line can be (a lot) faster on the CPU and varies widely on the GPU because on the quality and settings of the OpenGL drivers.

This is very true. Although, if you target GL, you'll batch the little line/widget drawing into buffers anyway, 1) because GLES forces it, and 2) for optimal performance. It's a little bit more code, but once it works, it renders really fast.
I'm a bit skeptical whether OO/LO needs GL acceleration on desktops, but with Wayland coming up and more and more UI moving GPU-side, it may be inevitable to port and take advantage of what's there already (GPU power).

3. Afaik the typical best solution is to combine the best of these worlds, which is also very sophisticated to implement.

Pain in the ass for coders, rarely good performance (context switching), but more importantly: a disaster for code maintenance. Think X11 bad.

Hence the easiest solution would be to do a OpenGL only solution but drop support for old GPUs because afaik the biggest OpenGL inconsistencies (problems) happen among old video cards (pre GL3).

Indeed. Or emulate GL on the CPU, and target only GL(ES) API. I don't know how software GL (Mesa or similar) would scale compared to whatever the current OO/LO UI rendering stack uses.

[wikinevick]

Minimal things like writing OOXML and stuff like that.

Incomplete and rather broken: I have documents that don't open in either suite. The german cities that sponsored most of that submitted the code to OpenOffice too but it was found to be poorly implemented.

[xeekei]

I must be lucky. I see people complaining about both LibreOffice's MS compatibility, and Firefox's performance and stability. Both applications never fail me.

[valeriodean]

You are completely missing the point on why the German cities are moving back to MS-Office (and Windows).

Are you speaking about the idiots of Freiburg?
About the competition between the two project, I think it is pretty much over:
1) LO can take the code from OO but not vice versa
2) LO has a more vibrant ecosystem, a lot more devs, a lot more involved company

[Aleve Sicofante]

Can someone please explain why the need for OpenGL in a word processor? It's a genuine question, no trolling.


Regarding Office compatibility, neither LO nor OO take it seriously. A much smaller project like WPS Office can get it right, so there's just not enough interest from LO/OO developers to do it. Even if they left older .doc formats aside and concentrated on OOXML, it would have a very serious impact in adoption of the open source suites. ODF formats are nice but they have failed to gain any traction, so why not adopt OOXML when it's fully open now, and both a de iure standard and a de facto standard?

[Filiprino]

Maybe this fixes some glitches that I have.
Could it be backported to Ubuntu 14.04 LTS?

As for Freiburg migration:

It seems there was a limited uptake of migration training, too. The result was an environment with both OpenOffice.org 3.2 and Office 2000 in use throughout the attempted migration.
Because Office 2000 did not support the OpenDocument format standard, this guaranteed a flow of documents in the formats used by both office suites, maximizing opportunities for incompatibility

Even newer versions of Microsoft Office can have incompatibilities with LibreOffice if you don't use ODF 1.2 or later. OO.org 3.2 clearly does not support the latest version of ODF. They didn't do the right choice by allowing people to use Microsoft Office. An even worse choice was to use such old versions of both office automation suites.

[pal666]

The fact is that humans are social by nature and you can be expected to open "another company's files", edit them, and send them back. And they must absolutely render fine on that other company's" leading commercial editor.

it works both ways. i.e. other company expects that they can open and edit your files and they must absolutely render fine on your libreoffice. or else they will dump msoffice and install libreoffice