Closed mtdvlpr closed 1 year ago
Hm.
Error: CompileError: WebAssembly.instantiate(): expected magic word 00 61 73 6d, found 3c 21 64 6f @+0
WebAssembly.instantiate(): expected magic word
Perhaps related to https://github.com/WebAssembly/spec/issues/1031
Duh. Ran the equivalent prebuild
commands.
I tested Malagasy and image zoom/pan. No issues on my end.
Check out my new commit, it improves the display of images when the aspect ratio isn't exactly 16:9.
MP3 to MP4 worked fine, ffmpeg download and ran the conversion. The progress bar didn't disappear, but not a major issue IMO.
MP3 to MP4 worked fine, ffmpeg download and ran the conversion. The progress bar didn't disappear, but not a major issue IMO.
This is what is displayed after the conversion finishes. No errors in console.
Strange that it doesn't go to 100%. After it reaches 100%, it will reset and dissapear
Found it. It doesn't call setProgress in the mp3 if branch
I'll test the following on Win32:
I'll test the following on Win32:
Awesome! Afterwards I think we're good to go with the release :) The Sentry issues that are still open can't be pinned down to a fix (yet), so I'll set them to resolved in next release.
For some reason, lff
isn't extracting properly on Win32. Other publications are, but not lff
. Hmm.
The JWPUB file gets downloaded, but not extracted.
M3 logs.-1667768249921.log
Are we possibly hitting some sort of memory issue on low-memory systems, since we're loadAsync
'ing the JWPUB files into memory before extracting them?
Might be related to this
I'll wait for the build to finish and try it out with the updated try/catches
Here's something interesting. The JWPUB file doesn't seem to be downloaded properly.
Ah okay. We're missing a check to see if the jwpub file is valid. No wait, we do check if the cached file size is equal to the file size from the api call
I deleted the 0kb JWPUB file to see what happens.
That would be in line with the stackoverflow answer:
I believe you would get this error after you stop your execution in between, i.e when the file is getting updated and this would corrupt the excel file.
Maybe because of the async execution the lff file is referenced multiple times and therefore download multiple times at the same time. You could look at the network tab to see if that's the case
Maybe because of the async execution the lff file is referenced multiple times and therefore download multiple times at the same time. You could look at the network tab to see if that's the case
Bingo.
"Both" files get downloaded at the same time, overwriting the same file. Zipper errors out on the first execution because the download from the second execution is still happening.
I can confirm that running the sync once the lff file has been fully downloaded (by a previous sync) results in a correctly unzipped lff folder and no further errors.
Makes sense that this is a new issue, seeing as pre-refactor, publications were always downloaded sequentially. If a pub already existed because it had been previously downloaded (no matter if it was 1 minute or 1 month ago), then the download was simply skipped.
This is definitely a scenario that would need to be accounted for, as I could see this causing errors in real-world usage, especially on newer installations that don't have publications like lff cached yet.
Would there be a way to modify downloadIfRequired
to:
If it has:
If it hasn't:
Working on it
see #634
It seems to work, because I'm now only seeing one call after clearing the cache and no errors
@sircharlo, can you check if it also works on win32?
I'm signing off. Time to go to sleep😴
@sircharlo, can you check if it also works on win32?
Negative, unfortunately.
Probably related to the low amount of memory of 32-bit systems (https://stackoverflow.com/questions/45920143/how-to-handle-array-buffer-allocation-failed-in-nodejs)
Edit: the error went away after relaunching the app, so I'm gonna call this a "you should really have more than 4GB of RAM to run media at meetings" situation.
Other than that, the repetitive-download behavior is in fact gone.
Slight regression introduced, see https://github.com/sircharlo/meeting-media-manager/pull/634#issuecomment-1304952468
Slight regression introduced, see #634 (comment)
Fixed in #635
I got my virtual Windows 32-bit machine working too! Everything seems to work, except converting mp3 to mp4. That goes wrong every time. The error is now caught and a human friendly warning is given. Since converting mp3 to mp4 is a real edge case, and it only goes wrong on (low-memory?) 32-bit machines, I'd say forget about it.
I got my virtual Windows 32-bit machine working too! Everything seems to work, except converting mp3 to mp4. That goes wrong every time. The error is now caught and a human friendly warning is given. Since converting mp3 to mp4 is a real edge case, and it only goes wrong on (low-memory?) 32-bit machines, I'd say forget about it.
Even worse, it works for me in Win32 😆 But agreed, not the end of the world.
Haha, then it's probably a memory thing.
I'd say we're good to go then
Will you do the honors, or me?
It's easier for you, since you can push directly to master. I'd have to create a pull request which I'm not sure has the desired outcome.
@sircharlo, so what happened is you wanted to push the release commit, but got blocked because you had to git pull first. Then when you pushed the merge commit got evaluated instead of the release commit, so no release workflow was executed. You'll have to do another chore(release)
@sircharlo, so what happened is you wanted to push the release commit, but got blocked because you had to git pull first. Then when you pushed the merge commit got evaluated instead of the release commit, so no release workflow was executed. You'll have to do another chore(release)
Yeah my bad, I redid it and it worked after. :)
Weirdly enough, the autoupdate didn't work for me until I deleted the %localappdata%\meeting-media-manager-updater
folder.
Hmm, it worked fine for me. The desktop shortcut also correctly updated this time.
Did you have a locally generated version installed by chance?
Nah, I didn't build on my machine at all. Must be a glitch with my setup, I wouldn't worry about it too much
Some things that need to be tested again before releasing: