GZDoom 3.3.1 Released

[Graf Zahl]

How many of you are working on this thing ? 3-4 ?

Not enough...

[Hipnotic Rogue]

Firstly, thanks for all your continuing hard work. I really do appreciate it.

Sadly, though, I think there may be a bug regarding music playback.

I have a ZIP file with a "music" folder in it that contains some tracks from Heretic II. I have used this as a music replacement pack for Heretic for some time now. The tracks are named correctly and has worked flawlessly until now.

I updated to 3.3.0 and now the engine only plays the default IWAD music, seemingly ignoring my zip file. I updated, today, to 3.3.1 and, sadly, its the same deal.

The ZIP file worked just fine up until and including 3.2.5. I'm using the x64 build on Windows 10 Pro.

Any help would be great.

[_mental_]

Is this music pack available somewhere? Without ability to reproduce the bug it can be a long guess game of fixing it.

[Graf Zahl]

First, check the console to see if it even gets loaded.
If that turns out not to be the case we indeed need the file.

[Hipnotic Rogue]

Thanks for the replies. I checked the console and it was certainly found and loaded by the engine.

Here's a link to the file:

http://www.mediafire.com/file/m16mei5vj ... ic+OST.zip

Hope this helps.

[Graf Zahl]

This doesn't load because it got no identifying marks to determine what is in there. The engine simply does not know if the ERITIC OST part should be removed or not. In such a case the loader does not alter the path. You have to move the 'music' folder one level up that it's in the zip's root.

[Hipnotic Rogue]

Thanks so much. I'm not sure how it worked fine in the past not, though. I've obviously had a brain-fart somewhere along the line...

Again, thanks for your help.

[phantombeta]

Those 6.5% are still too much do dump them, but they are low enough to stop working on that code. Currently I am trying to split up the renderer into platform independent code, GL4 code and GL 2 code and then deprecate the GL 2 stuff, never to be looked at or improved again. That separation is needed anyway if we really want to do a Vulkan backend so it isn't wasted work.

If someone volunteered to maintain the GL2 renderer when/if it's split (as in, fix bugs on it and such), would you accept the assistance?

[Graf Zahl]

It worked because the checking code was more lax. This, however, made it fail to load a few legitimate files, so I decided to err on the side of caution here.

[Major Cooke]

EDIT: Oh, no wonder, you posted just moments after he did.

[Graf Zahl]

Those 6.5% are still too much do dump them, but they are low enough to stop working on that code. Currently I am trying to split up the renderer into platform independent code, GL4 code and GL 2 code and then deprecate the GL 2 stuff, never to be looked at or improved again. That separation is needed anyway if we really want to do a Vulkan backend so it isn't wasted work.

If someone volunteered to maintain the GL2 renderer when/if it's split (as in, fix bugs on it and such), would you accept the assistance?

Sure. But don't expect this to last forever. The GL 2 renderer requires a vastly different approach to how things are done, compared to modern OpenGL and even more to Vulkan. So even if separated out, that's only a stop-gap measure to avoid tossing it right away. Long-term it's just not salvageable, especially when you consider that right now already more than 50% of users could run a Vulkan renderer. This is where things will inevitably go.
One day, when its user count shrinks too much chances are high to see it dumped very, very quickly.

[NightFright]

I have no objections regarding doing more of these hardware polls in future builds, btw. Always glad when you as a normal user can contribute to development somehow.

[Graf Zahl]

There surely will be other surveys in the future.

This one, I think, will not provide any new information if it runs longer. I'm probably going to do a news post about the results the weekend after next accompanied by a point release removing the survey code. The numbers we got were certainly interesting in multiple ways and provided important information about how to progress with future development - most importantly identifying a few things as having become obsolete (e.g. DirectDraw or Shader Model 1.4) because nobody or a single-digit number of users were running it.

[SiMpLeToNiUm]

Based on the numbers from the last survey, would you say people are receptive or hesitant to agree to disclose more detailed system information? How will this affect further development?

[Graf Zahl]

We got an engagement ratio of roughly 50%. That goes almost equally across all data points we can derive from other sources, e.g. the download counters for the latest GZDoom versions.
Of course the numbers do not tell us how people may react when being asked more detailed info.