{5} Accepted, Active Tickets by Owner (Full Description) (27 matches)

To reproduce:

1. open aegisub
2. open a video file
3. open an audio file that takes a few seconds to read into ram.
4. Before the audio finishes loading, either minimise the main aegisub window or switch to another workspace.

This results in what appears to be an infinite loop.

There are two backtraces attached. The first one was obtained with the steps described above, while the second one was obtained in slightly different circumstances: before loading the video, I loaded subtitles.

1) It's impossible to change the zoom level with the scroll wheel while hovering over the combobox unless I select a value from it using the left button. It doesn't matter if it's the same value as the current one. Just focusing the combobox is not enough. After selecting a value with the left button, the scroll wheel starts working. Scrolling a bit while hovering over the video will disable scrolling over the combobox again.

Perhaps scrolling over a moving widget shouldn't be possible at all...

2) Pressing enter in the combobox doesn't do anything, so using a custom zoom factor is not possible.

3) After changing the zoom level by scrolling over the video, changing the zoom level with the down arrow in the combobox has the funny effect of switching to the first level on the list - 12.5%. After that, the arrow keys behave normally (until the next time I scroll over the video). The up arrow will switch to the last level on the list - 300%.

I wonder if a simple textbox would be enough, instead of the combobox.

Aegisub revision is r6374. wxgtk version is 2.9.3. Operating system is arch linux (64 bit).

Addition to Changeset r2352:

Vector Clip tool should have a button to convert between \clip and \iclip.

The PCM WAV audio provider is not thread safe. It works most of the time, when it doesn't have to re-map the view of the file, but if two threads attempt to access parts of the audio that would lie in different parts of the file, a race condition can occur where one thread begins obtaining audio, reaches past the mapping check code and is ready/begins copying from the mapping. Simultaneously, another thread requests audio outside the current mapping, which causes the mapping code to tear down the current mapping and set up a new one. As a result, the copying in the first thread suddenly happens from an invalid memory location and the program crashes.

The only case where this can currently happen in Aegisub is the audio spectrum, since it is OpenMP multithreaded.

The proper solution is to allow the file mapping code to manage multiple mappings at a time and protect those mappings with multi-reader locks, possibly also adding some LRU logic to determine which mapping to tear down when the time comes.

This ties into #934, "Factor file mappings out of PCM reader".

Operating System name : Windows 7 English Version : Home Premium 32/64bit? 64Bit Any other relevant platform information. The Problem is in the picture attached.

To test, open any audio file, create a line with the same time as the start and end, then play it in the audio display.

aegisub revision: r6254 (2.1.9 branch) and 6282 (trunk). They're both affected.

Hi,

I found problem in SRT parser at one file. Probably it's due file is in UNICODE.

In attachment I sending that .srt file

As long I debug it at windows I found last blank line was read as "" empty sting, but dunno why text_line.IsNumber() at 524 line (case 5) subtitle_format_srt.cpp returned TRUE, so state was changed to 2 and in next read was end of file, so It crashed on Incomplete file.

I just wonder why empty line (String.Empty from C#) it took as number.

if I let be state 5 and just moved debug pointer to place same as will be if that IsNumber will be false then file was loaded, just there was two \N\N at last line

At linux version I getting error "Parsing SRT: Expected subtitle index at line 1", but there I don't have debugger, but mine guess will be somehow badly loaded unicode line from that file. Probably badly loaded unicode file and representing that UNICODE flag (0xFF 0xFE 0x31 0x00) as text.

Notice: After converting UNICODE --> UTF8 was file opened correctly

At linux I using 2.1.9 revision 6191, at windows 6186, but at 2.1.8 it was loaded fine

Regards Petr

Hi,

I've got last time randomly fall downs of aegi (it just disappeared) while I wanted paste (CTRl+V) lines from other script.

As long it's random I don't know how reproduce it, but I thing Backtrace could help find something.

Looks like it doing that when I do CTRL+V right after ALT+TAB with really short delay (As long I copying lines from other aegi window).

I using WX Widgets 2.8.10 (Since 2.9.x are pretty unstable with aegi at mine side)

Aegi is revision r6191 from 2.1.9 branch.

in Attachment I sending that backtrace, hopefully it helps.

--Petr

Allow audio loading/decompression to take place in a background thread, so you can still work on the subtitles while the audio is being loaded. The audio display should not appear before the audio was completely loaded, and all items in the Audio menu should be disabled. (Perhaps add a "Cancel audio loading" to it while loading.

To avoid interrupting/confusing the user, the audio display should not appear the instant the audio loading is finished, but instead the complition of the audio loading should be made clear in some other way, so the user can select when the audio display will appear. (Another possibility would be to first display it after a few seconds without input activity in Aegisub.)

An alternative to hiding the audio display while loading would be to display it immediately, and use it as a progress indicator similar to how Medusa does. All audio features would still be disabled while loading this way.

Currently it's possible to set the low cutoff frequency for the audio spectrum view, but the high cutoff is a fixed value, calculated from the samplerate.

It should be possible to set the high cutoff.

Also, currently the cutoff is specified in FFT frequency bands, it should probably be specified in Hz and have it transparently converted to frequency bands for display.

The enclosed patch adds accelerators for the "Recent" items in the "File", "Video" and "Audio" menus, "Export subtitles..." and "New Window" in the "File" menu. It also moves a few accelerators to more sensible place (in my opinion).

There's some reasonably robust handling of memory-mapped files in the PCM audio provider, but it's bound to the audio provider currently.

As this can be useful for other purposes it should be factored out to a base class of its own.

The numpad enter key doesn't commit changes when you just click a line from the subtitle list and perform some alteration with a keyboard shortcut. Only the main enter key works in this case. It does work, however, when you click on the text field for the selected line.

When using aegisub with the portaudio player on Ubuntu and probably other distributions that use pulseaudio + alsa mangling by default, the playback bar doesn't appear on the audio display. Attached is the output of aegisub compiled with -DPORTAUDIO_DEBUG.

The aegisub's tested revision was r3343, portaudio's was r1416 (HEAD).