Many crashes and corrupt files since 5.0.6 (character "ÿ" in files)

OK, it happened to you with the current developer version 551, right?

Yes 5.5.1 (783), 32b,

but with all versions of the last years after version 5.0.4 (543) 32 bit

For me, the ÿ always seems to appear on a line containing and open brace, {, in .c/.cpp source files.

I've also just seen this happen in seven places in a file named configure (no file extension) that doesn't have any highlighting. This happened while I was editing other files over the space of a couple of hours. This file has just been open in the background, and after seeing the first ÿ appear, I've been monitoring it for others.

Examples of lines changed:
"esac" was changed to "ÿ ac"
"# IFS" was changed to "ÿ IF"
"shift" to "ÿ if"
" q" to "ÿ ?"
"else" to "ÿ se"
"do :" to "ÿ :"

So doesn't seem to only occur at start/end of code blocks.

As far as I've seen, it's always the first characters within a line that are corrupted though. Has anyone seen it mid-line?

(PsPad is programmed in Delphi right? Is each line a separate WideString? I'd guess that as ÿ is 0x00ff in UTF-16. Knowing the string representation might give a clue as why the corruption occurs at the start of string.)

Highlight matching brackets is on
Completion of chars ( { [ < " ' is off
Code folding is off
Backup - switched off yesterday to see if it would make a difference

Version: 5.5.1 (783) 32b

I have been a PSPad user for many years. I observe the same problem that is described above. This example is from a C source file with more than 10000 lines (and there were other tabs open). After a save there were 3 lines (not contiguous) that suffered the "ÿ" problem, none of them were in the area I had done changes. To provide more detail the lines are shown in Hex Edit Mode:

Before: 0909736D4B6579546F6B656E293B090A
After: 0909736DFF007954203F3F00293B090A

Before: 09096F75742E726561736F6E3D72633B0A
After: 09096F75FF007265203F3F003D72633B0A

(The 3rd line I forgot to copy, but it was similar).

Nothing after the "ÿ" was displayed in the normal view, the "ÿ" showed at the 0xFF position. Note that just some of the characters were affected.
In all cases 5th and 6th characters were replaced with 0xFF00, and the 9th to 12th characters were replaced with 0x203F3F00.

Maybe these repeated replacement characters can provide some clou to solve the bug?

Ok, so combined with this bug (https://forum.pspad.com/read.php?4,78269,78278), it seems like 0x00ff is randomly being written to the heap.

I know this isn't an open-source project, but as you can't reproduce the problem, Jan, I'd like offer to try and debug it for you, if you would be happy to privately share the source/project. Hopefully should just be able to set a watchpoint on the strings on the heap to find out where the write is coming from.

Hi,
I just noticed this error using 5.5.1 (787) 32bit, Win 7, but unfortunately, I am not able to identify the exact step, where it appeared.

In a bash file an original list of filenames was extended to lines using regular expression replacement -
the original item "JT" should become:

./cmd/rnn-tagger-middle-high-german.sh ./vert_files/JT.txt > ./tagged_vert_files/JT_rnn_tagged.txt

The replacement worked on several hundreds of lines, but in this one line these places were replaced unexpectedly:

./cmd/rnn-tagger-middle-high-german.sh ./vert_files/ÿ .t绲.txt > ./tagged_vert_files/ÿ .t绲_rnn_tagged.txt

(not sure, whether they can be pasted via website).

hth,
vbr

Edited 1 time(s). Last edit at 2023-05-25 23:29 by vbr.

When will this error ever get fixed ??

I have this error of inserting character "ÿ" in files multiple times every day!

Coding the whole day with many big files open, search (and replace) in many big directories ...

Maybe, some users need ChatGPT interface etc. etc., but without a reliable editor this is all useless for me.

I use PSpad since many years, i donated some money, i check this out with every new beta version in 32 bit, but nothing happens.

After checking 5.5.1 (790) same procedure as before

If you don't plan to look after this error, it's ok for me, maybe there are only few (power) users with this problem.

But when there are no plans from your site, i should stop using PSpad,
not happy about that, but necessary for my development process.

Regards, Roger

heh... it won't *ever* get fixed... he's been beating his head against the walls for *years* over this bug, and doesn't have a clue...

Check out RJTextEd instead; it has all the features of PsPad, and doesn't have this bug.

is it that PSPad currently uses the Unicode editor instead of the old one fixed to one codepage?

are all modules (templates, highlighters, etc.) ready not to convert the codepage to the system one back and forth?

it seems like there is some place where it scrapes...

--
Vany
(PSPad 5.5.1.812 x32, W10h/p x64 en/cs)