Home Artists Posts Import Register

Content

This month has been an odd one as its not felt like it was productive since there were only 3 beta builds (9136, 9166 & last night's 9278) but on a quick review of those 3 changelogs & with an overall drop in the crash reports seen from the builds it has gone better than I thought :)

-

One of key things has been the non-legacy mode didn't explode in a burning fire of crashes as was something I was almost expecting & I probably shouldn't doubt myself so much with the extra testing that I've been trying to do before putting out new builds. So that is nice to now have something that can run albeit with some missing features without the need for the Winamp core program file (winamp.original).

Obviously there were a few things wrong (e.g. replaygain not working correctly causing the playback to seem like there's no audio) & the odd crashes but with the changes I've made to the crash report collection I'm getting more useful information when there is a crash to provide more certain fixes which we all benefit from.

-

Other things going on have been finally getting to the bottom of the weirder multi-threaded issues affecting the replacement local library plug-in & how it's handling live playcount update & related aspects of that. This is important getting that resolved as having the plug-in core stable it means I can now finally think about starting to work on the watch folders / external change monitoring as well as getting artwork display into the views.

Having those 2 features working comparable to the ml_local plug-in from 5.666 that the preview build currently uses is part of the block on there being a new preview build which is needed as build 7236 was ok for the time but I think it's now a clunky build to use compared to the recent 1.5.x beta builds.

-

Otherwise it's been about trying to refine things where possible, improving load times a bit more & with last night's build trying to work around runtime related issues that the modern skin engine (gen_ff) has which have been an on-going issue.

With gen_ff, it's main problem is that it doesn't always initialise things to a known state & when the WACUP core was being built with an older compiler it was easier for me to just generically patch the C runtime & it's memory & string methods to ensure any memory being allocated was pre-initialised (even if it's making things a little bit slower, it's better than things crashing) as well as avoiding some invalid parameter issues with the string methods.

I had tried back around August to have a 'legacy.dll' that would do the runtime patching but the anti-virus solutions didn't like it & it made things worse when there was a runtime conflict on some setups. What I've done now is the same approach but it swaps all of the memory allocation calls to use the newer compiler runtime I'm now using & that so far seems to be working (it helps once I'd resolved some initialisation weirdness that was preventing that patching being applied at the correct time so the old runtime methods couldn't be called).

The tl;dr on the above is that modern skins under WACUP should hopefully be more stable.

-

There's more in the build specific changelogs at https://getwacup.com/changelog/ but I think that's enough of my waffling for this update & I'll save doing a new preview changelog as it's now at 360 lines & that's maybe a bit too much to want to read though.

The main thing is that the WACUP builds seem to be moving in the right direction when it comes to stability & being able to consider offering a new public preview build once I've got more of the needed features in the replacement plug-ins & some of the WACUP core coded (which I should now get on & do!).

-dro

Comments

Anonymous

Thank you.