Enhancer Bug thread

@amigakit

Any reason why the AmiPDF bug fix didn't make it into this release?

@Raziel

Which bug fix are you refering to?

@broadblues

The very first post in this very thread

Bug in

MultiEdit 1.5 (28.05.2017)

Steps to reproduce: (Exact steps, otherwise this bug won't come up)

1) Start MultiEdit

2) (from the Menu or the button) open a file that is bigger in size than the scroll bar of MultiEdit's window
i.e. i'm using "System:Prefs/Env-Archive/contextmenus.cfg"
(Do NOT drag&drop it into the MultiEdit window, it needs to loaded from the menu, otherwise it won't come up)

3) After the file has been loaded, without clicking anywhere else, click on the "Search" button (it needs to be the icon, using the PD menu doesn't make the bug appear)

4) Type "exchange" in the Search field (pressing Enter afterwards makes no difference) and then "Find next"
The Search window will stay open, the Text will be scrolled up until "Exchange" will be visible and marked in the last line.

5) Again, without clicking anywhere else, click ONCE INSIDE the marked area that is now "Exchange".

MORE TESTS reveal that you should click exactly on the "c" in Exchange to triiger the bug 100%...it gets stranger by the minute...

The marked area will vanish and the cursor will appear.

6) Type ONE LETTER (i.e."r") and nothing else.


From now on the cursor will not be movable anymore with the keyboards cursor keys!(
UP, DOWN, RIGHT, LEFT won't trigger any movement)

Only if one either types another letter or mouse clicks somewhere else in the text it will become available again.

Very strange bug as i can only make it happen with these exact steps.

Dragging and dropping the file won't make it happen, using the PD menu for Searching won't make it happen, clicking somewhere else than inside the marked area won't make it happen, having a text file smaller than the scrollable area of MultiEdit won't make it happen.

Have fun chasing this one

P.S: There is a debug version of MultiEdit...where does it debug to and what?
Can i use the debug version with my bug to give more in formation?

I get occasional crashes with the new PTP file transfer utility on my X5000. The system just locks when it its nearing the end of loading the content on my iPhone 7.

By occasional I mean once every 3-5 times I use it. Especially the first couple of time I used it, it was bad and the crashes were almost every time. After about 20 tryes it is every 3-5 times.

I think it is the USB driver or something that crashes so I am no sure if it is PTP that causes the crash but below are the serial debug from two of the crashes.

@Raziel

Okay if that's the effect I thhink it is, that wasn't an AmiPDF bug, but a window.class one and is fixed at the OS level.

@Raziel

The Multiedit.debug contains sysmbls and debug info (stabs) , it does not dump any debug data, the only way it is of use if for gettig better stack traces when MEdit crashes.

I just discovered that the Enhancer 1.3 AmiPDF has a problem/bug that was fixed in the previous version. My account PDF statements don't have any text shown; just mostly blank pages. The XPDF window shows the message: "Couldn't create a font for UniversLTStd-light". It also shows similar messages for other UniversLTStd fonts. This is the same bug I had with the OS4.1FE AmiPDF. The Enhancer 1.3 AmiPDF has the same version number and date as the Enhancer 1.2 AmiPDF but is larger.

Apparently, the Enhancer 1.2 AmiPDF was doing better font substitution because my account PDF statements were readable and there was no error window poping up. OS4.1FE AmiGS also displays readable account PDF statements. Something went wrong with the new AmiPDF.

Here all all the font error messages:

Couldn't create a font for UniversLTStd-Light
Couldn't create a font for Univers_LT_Std_55
Couldn't create a font for Univers_LT_Std_45_LightBold

@xenic

See here for adding font mappings to the config file.

http://www.glyphandcog.com/manuals/XpdfImageExtractCOM/config.html

Add any mappings after the include line in the XPDF settings tab.

Why exactly the mappings aren't there by default I'm not sure, could be because it was built with a different libfreetype to fix the poorer rendering, in which case we there maybe a trade off between mappings and render quality.

PS I'd ask for sample but bank statements are a bit sensitive, do you have any other PDfs with the issue?

PPS Could you run snoopy while load sush a PDF? Looking at the src it's not obvious why it should fail in that particular way. I'd like to workout if it's an aliasing issue or a font comptabilty issue.

@broadblues
In the past I've spent hours trying to add fontmapping in the AmiPDF/Settings "Font map" tab. The entries I add are saved in ENVARC:AmiPDF/Settings but have no effect. I tried your method of adding the font mapping in the XPDF settings tab like this:

fontFile UniversLTStd-Light SYS:Fonts/_Ghostscript/n019003l.pfb

When I open AmiPDF I get this error in the error window:

Unknown config file command 'fontFile'

Either I'm doing something wrong or it just doesn't work.

I'll search for a PDF file that produces the same problem as my statements.

EDIT: I spent more than an hour searching for a PDF document with the same fonts and could not be read by the Enhancer 1.3 AmiPDF but could not find any documents on my system or on the Internet . I'll just have to revert to the previous Enhancer AmiPDF.

@Raziel
They also didn't fix the bug I reported regarding opening AmiPDF on another public screen. If I open AmiPDF on a public screen other that Workbench, AmiPDF momentarily brings the Workbench screen to the front before opening on the requested screen. Very ugly and unprofessional.

@xenic

Quote:

I remember looking for that issue and finding that it was caused by the DocViewer C++ class locking the screen before the AmiPDF subclass had processed the arguments, I couldn't see a way arround it at the time.

@broadblues
Quote:
I guess the DocViewer C++ class needs fixed? I tested the OS4FE AmiPDF and it appears to do the same thing; only faster and less noticable. The screen-switching is really noticable with all AmiPDF versions if you open AmiPDF on a screen with a different resolution than Workbench. I get a black screen for a second or 2 before AmiPDF opens on the requested screen.

I just got a user bug report for my USBSound program (at OS4depot) stating that it will no longer load and use external sound files specified in the program tooltypes. After extensive testing I've determined that USBSound works with the OS4FE sound.datatype installed but does not work with the Enhancer sound.datatype.

In addition, when my program gets the play time attribute using the SDTA_ReplayPeriod tagitem I get this result with the Enhancer sound.datatype:

Sound Seconds: 1600150964
Sound Microseconds: 0
which is absurd.

I get this result with the OS4FE sound.dtatype:
Sound Seconds: 0
Sound Microseconds: 324169
which makes sense for a short notification sound.

USBSound is a bit unusual in that it loads the sound files, closes the files and plays them from memory so that the files aren't locked indefinately. However, it worked fine with the 'official' OS4FE sound.datatype. I hate to disappoint the few users who might have USBSound installed, but I'm not inclined to attempt workarounds for the buggy sound.datatype that might then not work with the OS4FE sound.datatype.

The Enhancer sound.datatype is not backward compatable and needs to be fixed soon.

@Raziel
Thanks for reporting the problem with USBSound. For now, I think you'll have to reinstall the OS4FE sound.datatype (and disable the Enhancer datatypes that need it??) or be happy that the internal default notification sound stll works. It's such a small and relatively simple program that I never expected anything could go wrong (so much for assumptions .

@xenic

You mean you are using DTST_MEMORY rather than DTST_FILE?

MultiViewer is doing that and plays sound perfectly using the newer datatype. So it can't simply be that. MultiViewer is playing via the GUI though so that may make a difference if you are using the DTM_TRIGGER method to start it.

Can you provide a short code example which fails so I can make proper bug report for it?

@broadblues
Quote:
Yes. It optionally loads sound files (small ones) into memory and plays them for notification. The default sound is an internal array that was created with a bin2c program and works with DTST_MEMORY even when the loaded files fail.

Quote:
It's late in the day here and I'm beat, so I'll put together a code example in the morning and get back to you.

@broadblues
I got back to the code sooner than I thought. It turns out I misdiagnosed the problem. The problem is the SDTA_ReplayPeriod bug I mentioned in my earlier post. The program loads the sound file and checks the duration of the sound and rejects the sound if it's too long. I wanted to prevent the user from loading a rediculously long sound file for notification (like a 5 minute song). If I eliminate the play duration check, the loaded sounds play as they did with OS4FE sound.datatype.

There is still a bug to be fixed; it's just a different one that I originally thought.

Below is a minimal example program that you can compile and test with Enhancer sound.datatype and OS4FE sound datatype. Just change the path/filename in the code to an 8svx or aiff sound file on your computer. (wav won't play with OS4FE sound.datatype if you have Enhancer wav datatype installed)

You should find that the program prints a reasonable time with OS4FE sound.datatype and a huge time with Enhancer sound.datatype.

@xenic

Thanks for the xample code, bug confirmed here, I've made a mantis report and assigned it to the author of the datatype.

@broadblues
You could also add the Enhancer wave and mpega (assuming thats the datatype playing MP3 sound files). I was considering changing the USBSound code to calculate the sound play time from the SDTA_SamplesPerSec and SDTA_SampleLength attributes but the Enhancer wave and mpega datatypes return 0 for the sample length.

Also both the Enhancer and OS4FE sound.datatype return incorrect SDTA_SamplesPerSec, SDTA_BitsPerSample & SDTA_SampleLength for CDDA sounds files ripped from a music CD. They return 8 bit samples at 22050 samplerate and questionable sample length. You can include that it a bug report to AmigaKit and the OS4.1 bugzilla.

Here is an expanded version of the example program: