In Development or Planned
In Development!
The download will let you do configuration now,
but will cry "Not implemented yet!"
if you try anything else.
So, it cannot format paragraphs yet.
One thing is leading to another, again.
To let the Paragraph tool understand the most types of paragraphs,
it is necessary to make the assumption that their unit of
indentation equals TSE's TabWidth setting.
But different documents can have been created with different tab
widths.
Of course a user can manually adjust TSE's TabWidth to each
document themselves, or not and have less freedom in using
Paragraph.
But I want TSE's TabWidth automatically set for each document.
That should be an independent TSE extension, so I will look into
that before continuing with Paragraph.
(Heh, heh, heh: More hooks!)
Initial requirements:
-
Per-file support for one of two formatting types:
-
Text
- Paragraphs can be automatically recognized and formatted as the user types.
- A paragraph has no hard right margin. The window width serves as a variable one.
-
Program
- Formatting the current paragraph needs to be explicitly started by a menu or key.
- The formatting stops when the cursor leaves the paragraph.
-
Text
-
A file's "Program" or "Text" paragraph format type can be
- configured per extension,
- set per specific file.
-
Paragraph recognition for type "Text".
Doing so for all possible kinds of paragraph turns out to be impossible.
TSE is a plain text editor. The plain part means, that there are no invisible codes in the text, that mark where a paragraph begins and ends.
Therefore a paragraph needs to be detected based on the text's formatting.
These are factors that help to detect a paragraph:- Blank lines, and the beginning and end of a file.
- Indenting and outdenting.
- Bullet points.
My chosen solution is to only recognize paragraphs that adhere to certain rules.
For better and for worse, this does mean that existing paragraphs, that do not adhere to these rules, will be reformatted into different paragraphs!
A great insight was, that these rules could be made more flexibel by defining a left and right margin for each paragraph.
For the following new terminology is needed:- Here a "paragraph cluster" is a sequence of paragraphs that is delimited by a blank line or by the beginning or end of a file.
- Here a "paragraph cohort" is a sequence of paragraphs that share the same left margin.
-
A paragraph starts
- at the beginning of a file,
- or after a blank line,
- or with a bullet point,
- or with a line that is indented or outdented relative to the previous paragraph's left margin.
-
A paragraph ends
- at the end of a file,
- or before a blank line,
- or before a bullet point,
- or before a line that is indented or outdented relative to the current paragraph's left margin.
-
The first line of a paragraph cluster can be either
indented or not.
This is a user preference, and might need to be a configuration setting. - The first line of the first paragraph of a bullet point is not indented.
- The first line of a paragraph cohort, that is less indented then the previous paragraph cohort, is not indented.
- The first line of a paragraph cohort, that is more indented then the previous paragraph cohort, is double-indented.
- When a paragraph is formatted, its text is made to maximally fit the paragraph's left and right marging, with the exception of an indentation of its 1st line.
- For a "Text" file a paragraph's right margin is the screen width.
-
To do:
- Left margin loopholes.
- Tab sizes.
-
Paragraph recognition for type "Program":
-
To be defined.
I am thinking about 2 future solutions, to exist concurrently. Both will format only one paragraph of text after the user explicitly requested it.
Firstly to format a paragraph in a program comment, as defined by TSE's syntax hiliting definitions.
Secondly to for certain extensions format a paragraph between certain HTML-like tags.
-
To be defined.
-
In the future there might be an action to format all paragraphs
in a "Text" file.
This is too risky for "Program" files, because I have not been able to come up with a way to automatically recognize all their reformattable paragraphs reliably. -
TSE's word-wrap settings are ignored,
with the sometimes-exception of the right margin setting.
For a "Text" file the window's width makes for a practical right margin.
For a "Program" file TSE's right margin setting is more practical. For a "Program" file I also like the idea of setting the right margin to the maximum of TSE's right margin setting and the length of the longest line: This will initially keep the program's existing width intact, avoiding frivolous changes. -
Paragraph-compatibility with Word for "Text" files.
In using the Windows clipboard between Word and TSE, Word paragraphs are single lines in the clipboard. Therefore Word and TSE can exchange paragraphs, if TSE can toggle its paragraphs between being wrapped and being a single line.
Copying simple bullet points from Word to plain text seems implementable.
Copying plain text bullet points to Word might be doable with an extra manual action in Word to make Word recognize them as bullet points. To be tested.
TODO
-
MUST
- Implement "Text" formatting.
-
SHOULD (in descending order of priority)
-
Limited functionality to export/import "Text" files to/from Word (et al.)
I am thinking just paragraph-format and some bullet-points. -
If the Status macro exists, show a file's and paragraph's status.
Nerd alert: I will try to use the paragraph character, which has at least three different character codes depending on the user's relevant code page. -
Implement comment-based "Program" formatting.
This will use TSE's comment hiliting definitions to detect comments. -
Implement extension-based "Program" formatting, like text between tags.
My preliminary idea is to limit this to vertically delimited text lines, i.e. with the delimiting tags above and below them.
-
Limited functionality to export/import "Text" files to/from Word (et al.)
-
COULD
- Mayby make "Text" formatting not automatic, and introduce a new type "Auto[text]" which is automatic. It for example would allow extensions .txt and .text to respectively be formatted key-initiated and automatically.
-
Continuously hilite the current paragraph with a
configurable background color.
For example I like the subtle way in which the CursorLine extension can slightly modify the background color of the cursor line in GUI and Linux TSE. - A key directly to a menu to set the current file's format type?
-
WONT
- Solve the overly wide menu due to the wide key name field.
HISTORY
-
v0.0.0.5 3 May 2025 (2)
- Added cleaning up redundantly configured files.
-
v0.0.0.4 3 May 2025
- Added configuring "Program" and "Text" extensions.
-
v0.0.0.3 2 May 2025 (2)
- Fixed the bug, that configuring a key did not work if the menu was started with its configured key.
-
v0.0.0.2 2 May 2025
- In the menu merged the key definition line and the description line.
- Added ability to configure a key for the menu itself.
- It now remembers keys and a file's fortmat type across TSE sessions.
-
v0.0.0.1 28 Apr 2025
- Initial "in development" release. Just allows some ineffectual configuration. Has a new way to do key configuration.
This is a PROTOTYPE of the future tool.
eHelp displays an editable .ehelp file using TSE's help-engine.
Currently it cannot replace TSE's built-in help from within TSE, meaning it cannot provide help for TSE's menus, but it can be used in parallel to TSE's built-in help, for example by assigning a key to it.
It comes with tools to create a .ehelp copy of TSE's built-in help, to build an index, and/or to build your own TSE-like help for anything.
Run-time syntax:
ExecMacro("eHelp [ helpfile [ (sub)topic ] ]")
For example:
ExecMacro("eHelp tsehelp")
ExecMacro("eHelp tsehelp index")
ExecMacro('eHelp tsehelp "table of contents"')
ExecMacro("eHelp tsehelp execmacro")
There is no integration between any help files.
Typical use would be
- to provide TSE-like help for a tool,
- to provide an independent alternative to TSE's built-in help,
- to provide TSE-like help for anything else.
The tool has almost all features of TSE's built-in help, as well as some advantages regarding indexing and searching:
-
Index advantages:
- The index starts with a "first letter index".
- Topics and their subtopics are cross-linked.
-
Search advantages:
- A jumped-to subtopic or search result will be hilited.
- The "Search & View Topics" menu shows a "View Finds"-like list of search results and their topics, all of them with links!
- The help's "Back" function will also go back to the "Search & View Topics" list.
History: Recently the new tool's work name was changed from "sHelp" to "eHelp". In 2023 there was an old unfinished eHelp tool, which ambitiously tried to support multiple concurrent help files for TSE itself, the development of which was paused in October 2023, which will not be continued, and which is now replaced by this new much simpler eHelp tool, which will only support one independent help file at a time.
Remaining high-level steps:
- A minimum viable product (MVP) (The very basics, usable but limited).
- Versions with extra features.
Low-level steps towards an MVP:
-
Must:
- Several minor details to investigate and prioritize.
- More testing and bug fixing.
- Further documentation improvement.
-
Could, but I am leaning toward wont:
- Maybe at this stage already make eHelp capable of replacing TSE's built-in help?
- Maybe at this stage already include my own extended, up-to-date version of tsehelp?
v0.27.3 22 Apr 2025 (2)
Only updated the eHelp.s file.
Fully implemented the "Search & View Topics" menu option.
It now creates a "View Finds"-like list of search results
and their topics, all of them with links!
After selecting a link <Alt B> goes back to the list.
I have not done much testing yet.
v0.27.2 22 Apr 2025
Only updated the eHelp.s file.
Partially implemented the "Search & View Topics" menu option.
Search results are preceded by a line number and pairs of found
positions and lengths.
v0.27.1 16 Apr 2025
Only updated the eHelp.s file.
Added the help's search menu, but currently you can only select
the options for searching in the current topic.
v0.27 14 Apr 2025
Updated all 4 .s files.
Calling eHelp without parameters now gives a proper message.
Added the "Help on Help" topic to generating tsehelp.ehelp.
(In TSE's built-in help "Help on Help" is an exception:
It does not come from the .hlp files.)
Made eHelp_Hlp2eHelp call eHelp_CreateIndex after generating tsehelp.ehelp.
Made eHelp_CreateIndex save the newly-indexed .ehelp file.
Improved the documentation.
v0.26.3 10 Apr 2025
Only updated the eHelp.s file.
Fixed: A pop-up topic (like "Info->INTEGER") no longer shows the
"Info->" part in its window title.
v0.26.2 9 Apr 2025
Only updated the eHelp.s file.
Fixed: Using a down-key in an Index now works properly.
v0.26.1 8 Apr 2025
Only updated the eHelp.s file.
Partially fixed:
Using a down-key in an Index in some cases skipped a lot of lines.
Now it skips only one line.
Improved:
In the index: In subtopic-lines their topic is now justified against
the right side of the help window. It looks much better.
v0.26 6 Apr 2025
Added the eHelp_CreateIndex tool, which can (re)create an "Index" topic in an existing .ehelp file.
It differs in a few details from TSE's own Index:
-
All topics and subtopics are indexed.
At the top of eHelp's tsehelp's Index you will immediately notice some links to subtopics that TSE's own Index does not show. -
Subtopics are cross-indexed with their topics:
- They get their own entry with a parenthesized reference to their topic.
- They also get an indented entry below their topic.
-
The index starts with an index of the first letters of (sub)topics.
For example, to find "OneWindow" in the index, you can first select "O" in the letter index, and then scroll down.
Known minor issues:
- Using the arrow keys in the index can go too fast.
- The index does not show "()" behind functions.
v0.25.3 30 Mar 2025
Tiny documentation improvements.
Improved eHelp so that compiling it does not result in notes.
Set the debugging-check to FALSE to not waste performance.
Documented how TSE's debugger can be used on eHelp.
This tool shows a selectable font and its settings in a 16 x 16 character table.
The characters are shown as TSE shows them based on TSE settings.
The TSE session is temporarily shrunk to simplify a side-by-side comparison of different fonts and/or settings using 2 or more TSE sessions.
To do:
- Add font size as a selectable setting.
- Above the table, instead of the font and settings we chose, show the font and settings that Windows actually set based on our request. Or both.
9 Nov 2024 a user made a request, that I interpret as that eList should not always start at line 1, but instead on a list line based on the source's current line, like a search with the "v" option does.
v1.0.1.1
As requested eList now starts at the source file's current line.
When "zooming in" on a longer search string by typing a character, the
list's new current line is set at or as close as possible above the list's
previous current line.
It no longer uses the clipboard, which reduces memory usage,
increases speed, and leaves the clipboard intact.
v1.0.1.2
Disabled the main menu and help line during eList, which loses us nothing
and gains us an extra line of list space
Added numerous progress indicators, useful for processing big buffers.
Emptied the temporary helper buffers when finished to
improve memory usage.
v1.0.1.3
Two huge speed optimizations:
- A search will stop and restart when you type another character.
- When the search string is empty or matches an empty string, the source buffer itself is listed instead of making and listing a copy.
v1.0.1.4
Fixed the major bug in v1.0.1.3, that when eList had just been started and
the first typed character was <Escape>, then the current buffer became
empty.
While editing, the Git macro sporadically makes TSE ignore typed keys.
This is a very elusive and hard to confirm bug, because it almost never occurs.
My main suspect is the Git macro updating the git status
of the current buffer.
My suspicion is, that the macro's
Dos('git.exe … status -s …', _RUN_DETACHED_|…)
call makes TSE ignore keyboard input for a fraction of a second.
My first goal is to find a reliable way to reproduce the bug.
Aside, thinking ahead, I found this way out of the box
possible solution/work-around, that also might make Git usable
in the Console variant of TSE as well.
Currently Git "works" in the Console variant of TSE,
but with user-unfriendly updates of the current buffer's
git status.
7 Nov 2024 a user reported that Uniview sometimes only displays a too short part of the front of a line.
Tests confirm, that this typically happens for longer lines with lots of "Unicode characters".
A review of the Uniview extension makes me want to do a structural improvement of the implementation and documentation.
The upcoming eHelp project would also benefit from this, because the plan is to use UTF-8 for the new help file format.
Nota bene: This extension has no functionality yet, that is not already present in the Unicode extension.
Done:
- Identify a opened file's Unicode encoding if it has a byte order mark (which is rare).
- Simplisticly display the result with a message.
Todo:
- Identify an opened file's Unicode encoding or code page for all other cases.
- Display a pretty result somewhere.
- Give a signal when the Unicode encoding / the code page does not match the user's default.
- Make it configurable which signal types are used.
Research into 1-byte code pages suggests to me, that it is probably possible to in a useful way detect the three most used ones:
- Code page 437: Windows prompt, English (United States).
- Code page 850: Windows prompt, Other.
- Code page 1252: Windows GUI.
Code pages exist for Unicode too:
- Code page 65001: UTF-8
- Code page 1200: UTF-16LE
- Code page 1201: UTF-16BE
- Code page 12000: UTF-32LE
- Code page 12001: UTF-32BE
Tentative conclusion:
One separate "code page detector", that detects both Unicode
and the three most used 1-byte code pages, seems like a good plan.
This limited functionality should work in all TSE variants.
UniConv will be a newly written ASCII/CodePage/Unicode conversion tool.
It will have none of the interactive and responsive capabilities of the Unicode extension, and will purely focus on the conversion of character encodings.
If it can do a file-to-file conversions without loading a file into TSE, then it will be able to do much larger conversions, and users will be able to load much larger Unicode files. That does require a rewrite of the old code from Unicode.
Deprived from screen interaction, it would be nice if it could be started in these ways:
- From TSE as a macro to convert a string, (line?,) buffer, or file.
- From the command line as a macro.
- From the command line as an executable.
There will exist a critical macro load order dependency between the existing Unicode extension and the future BrowseMode extension below.
(Henceforth in short referred to as Unicode and BrowseMode.)
This requires Unicode to be updated to do its part in enforcing adherence to this dependency.
To be able to test this well, I am first going to try to finally fix the following technical debt.
Unicode is too big to be debugged in TSE's macro debugger.
(TSE's macro debugger starts but malfunctions for too big macros.)
Considered Unicode optimizations:
-
Obsolete code:
Unicode contains flawed, default disabled macro code for displaying the current character's real Unicode form in the the top-left corner of the screen.
This is a remnent from before UniView extisted.
Hopefully this can simply be deleted.
-
UniList:
Unicode contains macro code for displaying a list of all Unicode characters, for optionally selecting one, and for inserting it into the text.
Hopefully this can be split-off into a separate macro, perhaps named UniList.
-
UniClip:
Unicode contains macro code for copying, cutting and pasting TSE text to and from a Windows Unicode clipboard.
Hopefully this can be split-off into a separate macro, perhaps named UniClip.
-
UniConv:
Unicode contains macro code for displaying Unicode text in TSE, and for converting texts between ASCII, Windows-1252, and the major Unicode formats.
For a long time I have wished to split the conversion macro code off into a separate macro, perhaps named UniConv.
Hmm, now that I am thinking about this, there are a lot of arguments in favor of UniConv:- It would reduce Unicode's macro code.
- Its use could be shared by Unicode and UniClip.
- I already have other macros that currently use macro code copied from Unicode: They too could call UniConv instead.
- UniConv could be given the capability to convert "unlimited" file sizes.
-
TSE could edit much larger Unicode files for two reasons:
- Currently Unicode uses at least twice a file's size in memory for the conversion.
- If they mostly contain Western-European languages, then UTF-16 and UTF-32 files will shrink 2 and 4 times respectively when converted to TSE text.
23 Oct 2024 a user published their wish list for TSE, which included a wish for a "Write protected edit error".
I already needed to update my old "BrowsMod" extension, which helps to fulfill that wish.
I decided to completely rewrite my old "BrowsMod" extension into a new "BrowseMode" extension.
The old "BrowsMod" extension can still be found on Semware's "Classic" website.
The new "BrowseMode" will extend the old capabilities and add a new one.
The extended capabilities are, that it will automatically activate TSE's browse mode once for:
- Read-only files.
-
Buffers with names containing user-configured strings.
You can use this to automatically put an opened file in browse mode if its name contains a certain string.
For example, if a file exists in a certain directory, or on a certain drive or network share.
The new capability regards the following.
-
In browse mode TSE itself often already gives a warning
if you try to use a function that would change text,
for example duplicating a line or pasting a block.
However, in browse mode TSE does nothing if you simply type text.
The new extension can give you three types of signal for typing in browse mode: A sound, a shortly shown pop-up message, or a warning that needs to be acknowledged.
Status 28 Nov 2024:
- All stated functionality has been programmed.
-
I dare not publish a provisional version
until I have also added enforcement
of the macro load order of BrowseMode and Unicode.
A wrong macro load order opens the door for a not-attentive user to save a mutilated Unicode file.
For example, if BrowseMode sets the browse mode for a Unicode file before Unicode can convert it to TSE text, then a mutilated file is loaded into TSE, and it would be horrible if the user would save it. - Macro load order enforcement between BrowseMode and Unicode requires that Unicode will be improved too.
-
There will be the minor annoyance,
that executing BrowseMode via the "Execute..." menu
in order to configure it will also change the macro load
order of BrowseMode and Unicode.
This will disable BrowseMode until TSE is restarted, which will restore the macro load order.
A solution of sorts that I found, is to offer a public "BrowseMode -> setup" procedure, like the "cuamark" macro has, which can be accessed via TSE's "Macro -> Show Loaded" menu, and which does not require a TSE restart afterwards. - I want to add the browse mode optimization, that for a group of rapidly typed characters just one signal per configured signal-type is enough.
Linux TSE v4.50 rc 18 upwards (not expandable).
"DirList41" is the development version of the future "DirList" v4.1.
New so far:
-
In Windows DirList is now default independent of its environment's language settings.
This is achieved by retrieving link targets with Windows functions instead of dir commands.
This is a major step forward, even if it did create new loose ends.
Example loose ends:
- Link targets starting with weird prefixes.
- The Windows functions and the dir command each do not retrieve some of the other's details. The now default Windows functions do not retrieve some details of the link type. So I will still have to apply the dir command too for those languages for which the dir output can be interpreted. -
Gained a small but significant speed optimization by optimizing BigInt.
The access status column is now an access error column. Therefore when there is no error, "ok" is replaced by "-". When sorting this sorts all access errors together.
The "no link" indicator was changed from "." to "-". It subjectively looks better, and "-" is in my TSE WordSet, which helps accessing the column.
DirList's speed was increased again by only retrieving link targets for the link types that have them: junctions and symbolic links.
The --alllinktargets parameter generates the old list.
The --folowlinks parameter makes DirList list the child objects of directory links again. This is not recommended. -
Links are no longer followed.
This greatly reduces the run time and list size, and linked subdirectories are no longer counted multiple times in a common parent's directory size.
The following of links will later become a non-default option. - The link type column was condensed to be smaller.
- The column for TSE accessility is now a string indicating a reason.
- A link indicator (Linux) or type (Windows) column was added. Windows has a lot of link types.
- Windows: For junctions and symbolic links the link type and the link's target is now shown. It is shown between square brackets after the name, like "dir /al" does.
Still todo for DirList v4.1:
- Lots of (hopefully small) loose ends.
Math
I found my excuse for writing a Math macro that can do floating point operations: I will need it for the planned Picture macro to scale pictures.
To make optimal use of TSE's native capabilities and limitations, it makes sense to limit a floating point number to MAXSTRINGLEN, MAXLINELEN, or a TSE buffer size (255, 32000, or some value below 3.9 GB).
I cannot think of an excuse to make it longer than MAXSTRINGLEN, and the longer lengths would decrease calculation speed per digit. Tell me if you do know of such an excuse.
Alternatively, for others, if you do not mind installing new executables in Windows, then there is Eckhard Hillmann's FppPack_1_01.zip package on Semware's website. It was last updated on 12 Apr 2025.
From reading its documentation I gather that it provides high-level functionality in TSE, like a calculator-like expression-evaluator shell and a (block?) sum function, and that it provides basic math operators and functions by calling (closed-source) executables from the Windows command prompt.
This will be a tool and extension that lets us open a representation of a picture in TSE.
On 14 Dec 2024 I published XmasGift , an experiment in showing one specific picture in TSE.
A little more research showed how to show a representation of "any" picture in "any" version of any TSE variant.
The use of the word "representation" is very deliberate.
On the one hand Windows GUI TSE v4.2 upwards can typically show hundreds by hundreds of pixels in 16,777,216 colors.
On the other hand it might actually be even more useful what this
extension can do in a non-graphical Linux server/terminal,
in which Picture will at most be able to show a
picture as 80 by 25 pixels in 16 colors or less.
Even an impression of a picture could be real added value
in such an environment.
