K‐Lite Codec Pack Doesn’t Support Unicode?

View previous topic View next topic Go down

K‐Lite Codec Pack Doesn’t Support Unicode?

Post by Patrick Dark on Mon Sep 18, 2017 5:29 am

If, during installation of K‐Lite Mega Codec Pack, I change the hyphen‐minus (U+002D) character to a (more technically-correct) hyphen (U+2010) character, at the end of installation the K‐Lite Mega Codec Pack installer will indicate that three LAV codecs that it had just installed are missing and prompt me with a suggestion to delete references to the missing files. Furthermore, playback of videos will fail with a mention of graph data being missing. I’m guessing your software is not Unicode‐compliant.

Then, if I attempt to uninstall and reinstall K‐Lite Mega Codec Pack without changing the hyphens, the installer will tell me that files are in use and that a reboot is needed to complete installation at the end of the installation process (after I’ve customized installation options), but then no installation occurs after reboot. So I have to start the whole installation process all over again (re‐customizing the installation options).

Edit: I was using Windows 10, 64‐Bit.

Please fix.

Patrick Dark

Posts : 5
Join date : 2017-09-18

View user profile

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by notcyf on Mon Sep 18, 2017 6:35 am

Patrick Dark wrote:If, during installation of K‐Lite Mega Codec Pack, I change the hyphen‐minus (U+002D) character to a (more technically-correct) hyphen (U+2010) character, at the end of installation the K‐Lite Mega Codec Pack installer will indicate that three LAV codecs that it had just installed are missing and prompt me with a suggestion to delete references to the missing files. Furthermore, playback of videos will fail with a mention of graph data being missing. I’m guessing your software is not Unicode‐compliant.

Then, if I attempt to uninstall and reinstall K‐Lite Mega Codec Pack without changing the hyphens, the installer will tell me that files are in use and that a reboot is needed to complete installation at the end of the installation process (after I’ve customized installation options), but then no installation occurs after reboot. So I have to start the whole installation process all over again (re‐customizing the installation options).

Edit: I was using Windows 10, 64‐Bit.

Please fix.

The installer is probably not Unicode compliant, the software itself probably is though.

notcyf

Posts : 69
Join date : 2017-08-24

View user profile

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by notcyf on Mon Sep 18, 2017 6:35 am

Patrick Dark wrote:If, during installation of K‐Lite Mega Codec Pack, I change the hyphen‐minus (U+002D) character to a (more technically-correct) hyphen (U+2010) character, at the end of installation the K‐Lite Mega Codec Pack installer will indicate that three LAV codecs that it had just installed are missing and prompt me with a suggestion to delete references to the missing files. Furthermore, playback of videos will fail with a mention of graph data being missing. I’m guessing your software is not Unicode‐compliant.

Then, if I attempt to uninstall and reinstall K‐Lite Mega Codec Pack without changing the hyphens, the installer will tell me that files are in use and that a reboot is needed to complete installation at the end of the installation process (after I’ve customized installation options), but then no installation occurs after reboot. So I have to start the whole installation process all over again (re‐customizing the installation options).

Edit: I was using Windows 10, 64‐Bit.

Please fix.

The installer is probably not Unicode compliant, the software itself probably is though.

notcyf

Posts : 69
Join date : 2017-08-24

View user profile

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by Admin on Mon Sep 18, 2017 1:00 pm

I can reproduce the problem. The installer should support Unicode, so it is a bug somewhere. Investigating now.

Admin
Admin

Posts : 3619
Join date : 2011-06-17

View user profile http://codecs.forumotion.net

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by Admin on Mon Sep 18, 2017 1:27 pm

regsvr32.exe seems to replace the hyphen with a regular dash. This causes the file paths in the Registry to be different from the actual install path. I wonder why it does that, and if there are more such characters.

Admin
Admin

Posts : 3619
Join date : 2011-06-17

View user profile http://codecs.forumotion.net

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by Patrick Dark on Mon Sep 18, 2017 9:27 pm

Admin wrote:regsvr32.exe seems to replace the hyphen with a regular dash. This causes the file paths in the Registry to be different from the actual install path. I wonder why it does that, and if there are more such  characters.
It’s possible that regsvr32.exe is performing Unicode normalization along the lines of JavaScript’s String.prototype.normalize("NFKD") method; the formal Unicode decomposition of the hyphen (U+2010) character is the hyphen‐minus (U+002D) character.

One could test the theory by seeing if characters like Latin capital letter A with macron (“Ā”) (U+0100) decompose into the Latin capital letter A (U+0041) and combining macron (U+0304) characters—what Unicode calls a “canonical” decomposition—or the horizontal ellipsis (…) (U+2026) character decomposes into three full stop (.) (U+002D) characters—a “compat” decomposition like the hyphen case.

Patrick Dark

Posts : 5
Join date : 2017-09-18

View user profile

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by notcyf on Mon Sep 18, 2017 10:11 pm

Patrick Dark wrote:
Admin wrote:regsvr32.exe seems to replace the hyphen with a regular dash. This causes the file paths in the Registry to be different from the actual install path. I wonder why it does that, and if there are more such  characters.
It’s possible that regsvr32.exe is performing Unicode normalization along the lines of JavaScript’s String.prototype.normalize("NFKD") method; the formal Unicode decomposition of the hyphen (U+2010) character is the hyphen‐minus (U+002D) character.

One could test the theory by seeing if characters like Latin capital letter A with macron (“Ā”) (U+0100) decompose into the Latin capital letter A (U+0041) and combining macron (U+0304) characters—what Unicode calls a “canonical” decomposition—or the horizontal ellipsis (…) (U+2026) character decomposes into three full stop (.) (U+002D) characters—a “compat” decomposition like the hyphen case.

normaliz.dll(C:/Windows/System32) does this, and is used by explorer.exe, not sure about regsvr32.exe

notcyf

Posts : 69
Join date : 2017-08-24

View user profile

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by Admin on Mon Sep 18, 2017 10:33 pm

I have checked and A with macron is also changed, so this string normalization is being performed. This means I will have to disallow using such characters.

Admin
Admin

Posts : 3619
Join date : 2011-06-17

View user profile http://codecs.forumotion.net

Back to top Go down

Re: K‐Lite Codec Pack Doesn’t Support Unicode?

Post by Patrick Dark on Tue Sep 19, 2017 7:43 am

Admin wrote:I have checked and A with macron is also changed, so this string normalization is being performed. This means I will have to disallow using such characters.
If it helps, there’s a *.txt file that could be parsed at https://unicode.org/Public/UCD/latest/ucd/DerivedNormalizationProps.txt.

Patrick Dark

Posts : 5
Join date : 2017-09-18

View user profile

Back to top Go down

View previous topic View next topic Back to top


 
Permissions in this forum:
You cannot reply to topics in this forum