Unify INI values by hansu · Pull Request #3704 · LinuxCNC/linuxcnc

hansu · 2026-01-17T08:57:01Z

To be done

Adapt Gmoccapy - change spindle => spindle_0 in *.ini files #3488
Update sample configs for DEFAULT_LINEAR_VELOCITY [TRAJ] -> [DISPLAY]
Increase INI file version and set up update-ini script

zz912 · 2026-01-17T15:33:50Z

* [ ]  Adapt [Gmoccapy - change spindle => spindle_0 in *.ini files #3488](https://github.com/LinuxCNC/linuxcnc/pull/3488)

Feel free remake (or close) #3488 how you need. Git is not my friend, so I cant help you.

#3488 also does:

INI file from PNCconf is more verbose in build_INI.py
correct MACHINE in 7_axis.ini
correct MACHINE in 6_axis.ini

andypugh · 2026-02-04T13:35:06Z

update_ini is designed to be scriptable, (-d and -f flags) and I think I had a meta-script that ran it on all the sample configs.
I probably can't find it any more, but I could have a look.

hansu · 2026-02-04T14:28:02Z

update_ini is designed to be scriptable, (-d and -f flags) and I think I had a meta-script that ran it on all the sample configs. I probably can't find it any more, but I could have a look.

Yes that would be helpful. I only used tested with the -d option so far.
The renaming stuff is done so far, only some parts which moves or copy some values are missing.
And a bit refactoring maybe.
Also a little bug where it generates two [KINS] sections...

andypugh · 2026-02-04T23:58:20Z

Yes that would be helpful.

Sorry, I looked and it's nowhere obvious. It might be sat in a git stash somewhere, but I think it's probably been lost.

I doubt it was anything special, I am not much of a scripting wizard.

hansu · 2026-02-09T00:13:44Z

I copied now

[TRAJ]DEFAULT_LINEAR_VELOCITY -> [DISPLAY]DEFAULT_LINEAR_VELOCITY
[TRAJ]MIN_LINEAR_VELOCITY -> [DISPLAY]MIN_LINEAR_VELOCITY

The values are/will not be not used in TRAJ any more. Do you think we should delete them from TRAJ or keep them as a precaution ?

c-morley · 2026-02-09T03:51:55Z

Ate you sure they ate not used in the trajectory planner?
Iirc they are used in non trivial machines.

hansu · 2026-02-09T21:18:02Z

Ate you sure they ate not used in the trajectory planner? Iirc they are used in non trivial machines.

DEFAULT_LINEAR_VELOCITY I overlooked, but MIN_LINEAR_VELOCITY is couldn't find in the trajectory planner.

c-morley · 2026-02-09T21:46:40Z

Ahh yes DEFAULT is what I was recalling.

hansu · 2026-02-11T01:29:04Z

I think the script is ready for the first review.

@andypugh How do we handle the message boxes if upgrading from <1.0 to 1.2?
Is it too irritating when there are multiple LinuxCNC versions in the list like this?

zz912 · 2026-02-22T13:40:05Z

Hi Hans, when do you expect to finish this PR? I'm currently converting the spindle control from PWM to hm2_modbus. So I could test if there are any bugs when changing SPINDLE => SPINDLE_0.

hansu · 2026-02-22T13:51:35Z

It's more or less ready from my side. This is waiting for a review and then all the INI files needs to be converted by the update_ini script and committed.
But this will be done just before 2.10 is being released I guess.

zz912 · 2026-02-22T14:01:50Z

I asked the wrong question.
When do you expect to merge this PR to master?
I'm currently working on hm2_modbus communication.
If it were merged in a few days. I'd wait for that.

I'm just asking. I don't want to do it twice.

hansu · 2026-02-22T15:21:00Z

I think it will take some weeks

zz912 · 2026-02-22T15:43:28Z

Ok, thanks for information.

andypugh · 2026-02-23T11:57:48Z

I think that the very long message box is fine. We really want users to be aware that a lot has changed, and to think before pressing the buttons.

BsAtHome · 2026-04-07T17:06:24Z

How does the update script handle #INCLUDE syntax?

The include handling is expanded by the linuxcnc script and you don't really see the whole truth when running. Well, until now, because the new INI parser does the include internally and does not rely on external merging.

But still, the update script should be aware of #INCLUDE directives and act accordingly. Maybe there needs to be an additional interface in the new INI parser to extract filename/linenr information from python (already possible in C++).

BsAtHome · 2026-04-18T16:21:45Z

The update_ini program also needs to update quoted strings (see #3946 and #3947).

The new ini-parser in python allows you to recall the exact file and line of any variable by using linuxcnc.ini.lineof(section, variable [,num]) (returns a tuple). This allows you to read the old ini-file as text, converted into an array of lines, and then change any line accordingly. The reason to use this feature is because #INCLUDE statements skew the program's view of the content, but the original source of variables can still be recovered by using lineof().
A small problem is line continuation, which are not recoverable from the parsed value (was merged into one value). However, I don't think that lines with continuations are lines that generally need change. But, it is possible to look at the previous line in the source text and see if that had a continuation and act accordingly (lineof returns the last line of the continued value).

…qt_istat

There is no MIN_LINEAR_VELOCITY in the TRAJ section.

…nuxCNC#3497)

… (TRAJ --> DISPLAY)

…tted

BsAtHome · 2026-06-14T18:35:37Z

 # We want to work with the base INI file here, not the expanded version if #include is used
 filename = re.sub(r'\.expanded', '', filename)



The "expanded" version is no longer generated. If you find an expanded inifile, then it can be removed. The Tcl code has been fixed to use the new inifile parser.

BsAtHome · 2026-06-14T18:40:18Z

+included_files.append(filename)
+pattern = re.compile(r"^#INCLUDE[ \t]+(\S+)")
+with open(filename, "r", encoding="utf-8") as f:


The include match is not entirely correct, although it should normally not matter I guess. The parser uses ^#INCLUDE[ \t]\s*(.*)\s*, which captures anything after the statement and strips leading/trailing whitespace.

However, what may be wrong is that opening the included file uses a specific path methodology with the following cases:

path/file, if $current_include_filename has a '/' in it:

--> $(dirname "$current_include_filename")/path/file

else --> use open(path/file)

~/path/file --> $HOME/path/file

/path/file --> absolute

You need to devise your own running view through the tree of files while tracking through the sections and variables. Includes can happen anywhere, cannot recurse (beware of loops), but may be included multiple times. You must process each file in sequence as the parser would read the files and convert in sequence. You also need tracking already converted files to not double convert and assure that a converted file will be valid when read again in another place (perfectly legal situation).

You also need tracking already converted files to not double convert and assure that a converted file will be valid when read again in another place (perfectly legal situation).

Double conversion should be no problem - if a file is already converted then there is simply nothing to convert.

The include match is not entirely correct, although it should normally not matter I guess. The parser uses ^#INCLUDE[ \t]\s*(.*)\s*, which captures anything after the statement and strips leading/trailing whitespace.

But this would include spaces in the filename:

Yes, it can include spaces. That is how the original also worked. The behaviour was taken from that work.

BsAtHome · 2026-06-14T19:11:37Z

+
+for current_file in included_files:
+    print(f'===> Updating "{current_file}"...')
+    if version == "$Revision$" or version < "1.0":


I don't think it would be a good idea to handle included files separately because there is no guarantee there is anything in them you can understand without the context of where you are in the ini-file sequence.

As mentioned above you must read through and convert the files in order to be able to track where you are.

If you don't want to merge all the includes together, you have to handle them separately.

Well, yes, you need to handle them one-by-one. However, you must do so as-if you are handling one file and treat everything as one stream while tracking the actual origin.

Example: include variables from a file:

[Tag] #INCLUDE otherfile.inc [OtherTag] #INCLUDE otherfile.inc

and otherfile.inc contains:

# Note: a nested include is perfectly fine #INCLUDE common.inc ValidInTag=blabla # Changing the following tag will become invalid after change but is necessary for [Tag] BecomesInvalidInOtherTag=nono

Although the error scenario may be rare, the way these includes are made is not. The above include sequence is f.ex. seen in axes and joints that share most parameters. If a converted include file is re-read and would require changes, then you have to flag an error in auto-conversion and say it has to be done manually.

It would also be perfectly legal for an include file to end with a new section header. Returning one level up would then add variables to the section the include file ended with. This would be really borderline sane, but there is nothing that prevents it and the parser happily accepts it. Such scenario will automatically work when you treat the whole ini-file as a text-stream.

The "easy" way out for all the possible problems is to drop auto-conversion of INI-files that contain includes. However, that would make it harder for users. But when you do convert them, then you must support the whole structure and deal with the complexity of how the parser sees and reads the INI-file data when read.

Yeah that's a bit tricky. But I don't think it's worth the work to handle this case.

But feel free to join this game...

… section at all

BsAtHome · 2026-06-15T10:29:55Z

It is good that you run the update/test after the ini-file is tested for validity:

linuxcnc/scripts/linuxcnc.in

Lines 417 to 422 in c1f4b6c

    
           # Test whether we have a valid INI-file that can be parsed without error. 
        
           # Running inivalue without section/variable filter should return the value of 
        
           # the first variable in the first section of the file. If that fails, there is 
        
           # a parse error in the file or it is empty. 
        
           inivalue "$INIFILE" > /dev/null || { echo "E: The INI-file contains errors that need to be fixed."; exit 1; }

It guarantees you already know that the file conforms to the formal standard. The content is then of only concern.

BsAtHome · 2026-06-15T17:08:08Z

Alright then... have a look at this: process_ini.py

It takes an ini file and processes it as one stream white tracking includes and converting recursively. Two functions are provided to handle conversion (handleSection and handleVariable)

The only limitation is that it assumes that you do not re-read an include file that has sections in it. This case would already produce a warning in the new ini-parser (same section name gets its variables merged), so it is probably safe to assume this will not happen.

hansu · 2026-06-15T20:57:17Z

But how should that fit in the current update_ini script?

BsAtHome · 2026-06-15T21:13:55Z

But how should that fit in the current update_ini script?

With care? ;-)

But seriously, it needs some more work, yes. This was just to show that making a full stream editor for the ini-file is not that difficult. Will make some "suggestion" when I'm done hacking the script.

hansu added the 2.10-candidate would be nice to have fixed in 2.10 label Jan 17, 2026

hansu force-pushed the unify-ini-values branch from 26ef94e to 75b7af8 Compare January 27, 2026 21:26

hansu mentioned this pull request Jan 27, 2026

Gmoccapy - duplicate spindle speed settings in *.pref and *.ini files #3497

Closed

hansu force-pushed the unify-ini-values branch 2 times, most recently from 619faee to 0932268 Compare February 2, 2026 22:08

hansu force-pushed the unify-ini-values branch 2 times, most recently from ef0fb47 to 9cc5bac Compare February 4, 2026 14:20

hansu force-pushed the unify-ini-values branch from 1d4850e to c7afc84 Compare February 9, 2026 21:15

hansu force-pushed the unify-ini-values branch from bfb59a9 to e4c8171 Compare February 11, 2026 08:18

hansu requested a review from andypugh February 11, 2026 18:25

hansu marked this pull request as ready for review February 22, 2026 13:44

BsAtHome mentioned this pull request Apr 18, 2026

ini: do not treat quotes as a special character in values #3947

Merged

hansu force-pushed the unify-ini-values branch from 50c6463 to dc704c6 Compare April 18, 2026 18:33

hansu added 19 commits June 10, 2026 20:59

gmoccapy: use now ..._SPINDLE_0_... instead of ..._SPINDLE_... using …

2d32ca5

…qt_istat

gmoccapy: use MAX_FEED_OVERRIDE using qt_istat

bbbafdb

gmoccapy: {MIN, MAX, DEFAULT}_ANGULAR_JOG_VEL from qt_istat

bc866a1

sample-configs/axis: move MIN_LINEAR_VELOCITY from TRAJ to DISPLAY

3a5d4f2

There is no MIN_LINEAR_VELOCITY in the TRAJ section.

docs: remove [TRAJ] MIN_VELOCITY

424723a

gmoccapy: use spindle min, max and default directly from INI file (Li…

8f5b01a

…nuxCNC#3497)

gscreen: use DEFAULT_LINEAR_VELOCITY from DISPLAY instead TRAJ

100ff69

gmoccapy: DEFAULT_LINEAR_VELOCITY + MAX_LINEAR_VELOCITY from qt_istat…

c7075a1

… (TRAJ --> DISPLAY)

gmoccapy: change SPINDLE => SPINDLE_0 in *.ini files

11dfab3

update_ini: basic steps for updating 1.1 -> 1.2

8329192

update_ini: fix keep emtpy lines at the end of a section

6bce89e

update_ini: fix/improve regex for finding the DISPLAY section

5259950

update_ini: copy values from TRAJ to DISPLAY

4a70de4

update_ini: remove MIN_LINEAR_VELOCITY from TRAJ

b3b20e3

update_ini: fix \n to work also from upgrading <1.0

0c7a3c3

update_ini: show info for each version step in one message box

d9bc65b

update_ini: remove warning if no changes in HAL files

d8d5904

docs: update "updating to 2.10"

1928c17

update_ini: get the included files in the INI file

7e24a91

hansu force-pushed the unify-ini-values branch from dc704c6 to 7e24a91 Compare June 10, 2026 21:54

hansu added 3 commits June 14, 2026 17:29

update_ini -includes: going through all included

01b9084

update_ini -includes: Copy section only if exists

4bc0907

update_ini - includes: [EMC] Match also until EOF if INI file is spli…

9c4cd69

…tted

BsAtHome reviewed Jun 14, 2026

View reviewed changes

hansu added 2 commits June 15, 2026 12:09

update_ini: don't handle .ini.expanded - this is not generated any more

6825900

update_ini -includes: copy everything from a file also if there is no…

bc723a5

… section at all

		# We want to work with the base INI file here, not the expanded version if #include is used
		filename = re.sub(r'\.expanded', '', filename)

Conversation

hansu commented Jan 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zz912 commented Jan 17, 2026

Uh oh!

andypugh commented Feb 4, 2026

Uh oh!

hansu commented Feb 4, 2026

Uh oh!

andypugh commented Feb 4, 2026

Uh oh!

hansu commented Feb 9, 2026

Uh oh!

c-morley commented Feb 9, 2026

Uh oh!

hansu commented Feb 9, 2026

Uh oh!

c-morley commented Feb 9, 2026

Uh oh!

hansu commented Feb 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zz912 commented Feb 22, 2026

Uh oh!

hansu commented Feb 22, 2026

Uh oh!

zz912 commented Feb 22, 2026

Uh oh!

hansu commented Feb 22, 2026

Uh oh!

zz912 commented Feb 22, 2026

Uh oh!

andypugh commented Feb 23, 2026

Uh oh!

BsAtHome commented Apr 7, 2026

Uh oh!

BsAtHome commented Apr 18, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

BsAtHome commented Jun 15, 2026

Uh oh!

BsAtHome commented Jun 15, 2026

Uh oh!

hansu commented Jun 15, 2026

Uh oh!

BsAtHome commented Jun 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

hansu commented Jan 17, 2026 •

edited

Loading

hansu commented Feb 11, 2026 •

edited

Loading