|
Post by George on Apr 11, 2024 15:24:03 GMT -5
MUEH: the +Std is weirder. I'm changing the color validation code as it looks like there's some checks that don't need to be done with the new parser, and are now causing problems.
I'll post a new version tomorrow.
Interesting problems.
George
|
|
|
Post by George on Apr 11, 2024 14:49:59 GMT -5
MUEH: Thanks for all the testing, it's invaluable.
First - the 2nd bug. I think this is actually a bug in the production version, which the new version has plugged.
SHOW syntax is describes as:
SHOW [ search-string ] [ start-column [ end-column ] ] [ FIRST | LAST | NEXT | PREV | ALL ] [ C ] [ Q ] [ T ] [ PREFIX | SUFFIX | WORD | CHAR ] [ line-control-range ] [ U | NU ] [ color-selection-criteria ] <============ [ TOP ]
Which says 'color-selection-criteria' and that means only LIME, not +LIME. i.e. search only, not change.
I think here you want the FIND command, which would allow the +LIME operand.
I know, it used to work, but the revisions in this case are just enforcing what the doc. describes. The previous versions actually had a bug.
George
|
|
|
Post by George on Apr 10, 2024 14:15:54 GMT -5
MUEH: This is showing that converting the 'rules' for every command into definitions for the new parse code is much harder than I thought. Kwds that are not defined, by default, become literal operands. Hard to believe (well maybe it isn't) that stuff like this escaped my testing.
BTW, just a comment, about Robert's comments that this is an untested, maybe unreliable release.
Yes, it is not fully tested, nothing with SPFLite's complexity and interaction of various options could ever be. Even with some automated testing tool for which he has been pushing. (And which I doubt could ever be created)
But I have been using this new release for editing the SPFLite source code itself, and have never said "d*mn" it messed up.
I think the fix for this latest bug will have to wait till tomorrow. I'll have to do a full review of the parsing rules to see if I've messed up some other definitions.
George
OK - 24102 is available.
|
|
|
Post by George on Apr 10, 2024 11:47:53 GMT -5
MUEH: Ack! Of course! Dumb stuff like this is my downfall, the brain ain't what it used to be.
George
P.S. I posted a 24101A version with the fix. The model definition was corrected, it wasn't a coding error, but still a bug.
George
|
|
|
Post by George on Apr 10, 2024 10:09:35 GMT -5
Jo: No problem reporting it here at all. During my V3 testing I ran into many bugs of course, and surprisingly, many existed in the production version so were not part of the V3 changes.
So is that last file simply not visible because it's on the next page? Or was it Excluded or Forgotten?
George
|
|
|
Post by George on Apr 10, 2024 9:54:13 GMT -5
Robert: You're probably right on lots of your points, but I do object mostly to being called lazy. I've been working on this almost full time for nearly 4 months. Why? Because I was tired of the state of the code. And I'm tired of being the only person in the world who has any inclination to actually work on SPFLite.
BTW The Doc. in this release has been updated, even your latest request.
I've spent weeks testing this release, it may still have bugs, but it is NOT untested. Saying I'm foisting an untested program on the users is a bit much.
As to removing unused stuff. Do I expect anyone to bemoan the loss of Numbering support? Fold? Command chaining? Get real. I didn't drop them from laziness, individual commands of their nature were trivial to do. It just galled me because they were basically useless. And Command Chaining" - well with 8 pages of dense text in the Help file to describe it, nobody would ever put the effort in to learn how to use it. Sort of like your format change support that we yanked.
You are back at harping to me on stuff like a Design Document, reviewing it with users, yourself etc.; setting up testing plans and automation. Just how big a development team do you think works on this?
Yes, your suggestions and involvement are certainly less than in the past. Why? Because every positive suggestion is wrapped in negative comments about my work style. I don't need to be told that year after year. If you wanted to have things done to your high standards, you should have continued on the QSPF path and then you could sit back and snigger at SPFLite.
I know I've been talking shutting this down for a long time, but I love doing this, it's really hard to just stop. But this time it really has to end. This will be the last significant change. I want to leave the source in as clean a state as I can manage. My peripheral neuropathy has basically killed my legs and has now jumped to my hands; I can no longer do up buttons; I can't tie my own shoes; and typing is more Bksp than forward.
Until somebody steps up to take over SPFLite, and fix everything you say is wrong with it, everyone is stuck with this 'unproven, untested and untestable alpha program' that I've turned out.
George
|
|
|
Post by George on Apr 10, 2024 9:00:30 GMT -5
MUEH: Embarassing! That was one of the first things I ran into with Macros, there was no parsing model for them. I spent much more time trying to get Stefan's LP macro to run properly. I used it because it's the most complicated macro I have. Spent days on that one.
I've corrected this, loop for an updated EXE later.
Jo: I can't seem to get a failure. I've gone through FF logic, and file open logic, and made a couple 'tweaks' where some error conditions were not quite right.
Did your FF trigger any popups due to missing Profiles, or anything else odd? I'm going to post an update (24101) regardless. We'll have to see if it is still a problem.
George
|
|
|
Post by George on Apr 9, 2024 12:41:04 GMT -5
Your Help is needed. For the last few months I have been working at cleaning up SPFLite code. Mainly this has revolved around replacing the command parser and making accompanying changes to the generalized search routine. As well, some old commands and functions have been removed, such as: - Support for the FOLD option
- Support for source code numbering (the commands AUTONUM, NUMBER, NONUMBER, NUMTYPE, ORDER and UNNUMBER)
- Support for "Command Chaining". This is NOT the ability to separate commands on the command line with a ; character.
All together, this was an extensive set of changes. I have spent several weeks now in testing and checking out all the various commands. But there is simply no way I could declare it 'clean'. I am hoping this is where some of you can help, by simply using the new release. (Being cautious of course) and reporting any differences / bugs / etc. SPFLite3======== I've decided to bump the release from 2 to 3 and keep it separate from the current production release - SPFLite2. It will install in a separate folder from SPFLite2. And both versions can exist together. However, there is still a common point: the HomeFolder (\Documents\SPFLite\) which contains, normally, the CFG file and sub-folders like MACROS etc. To ensure 'a way back', early Alpha versions of SPFLite3 will automatically create a backup copy of the CFG file, with a date/time stamp added, in the Homefolder. I have been sharing the CFG file throughout testing and have not experienced any problems. Note: If you use the CFGMaint utility, the Release 3 version will report the deleted CFG entries and not export them. If you do not IMPORT a Release 3 Export, everything will be fine. If you do, and then run SPFLite2, these entries will be recreated with default values. I seriously doubt anyone will be impacted by this unless you are actually using the removed feature support. The install of SPFLite3 is performed the same as SPFLite2, it is a full installer package. Since the download is too large to be embedded in a Forum post, the following link will download it from Google Drive. (You do not need a Google ID to download it) Installing
The SPFLite3 Installer need only be downloaded and installed once, it will create the new SPFLite3 install folder SPFLite3 Installer
If you have run the above installer, updated versions can be downloaded below. The new EXE file should then simply replace the previous version in the install folder. Changes
* Corrections to some command parse table entries. * More corrections to the command parse tables * Correct an old production release bug which prevented the File Manager FF (Find in Files) command from processing files which required usage of the XFORM macro support to read the file. * With 24104, I've corrected the WDIR command parse model, and removed the backup of the CFG file at every startup. * Corrected the File Manager FF command, it was not always switching back to FM Mode after the search * Correct missing parse model for ACTION, and correct the handing of line references in the SORT command. * Correct a parse error in the SORT command. Alter CHANGE ? from displaying the Help for CHANGE, to displaying the CHANGE CS/DS setting. * Another correction to SORT, it was not fully resetting search parameters from a previous command. * Correct an old bug from the production version. Creating a new INSTANCE was not copying the correct previous instance, it was always using DEFAULT. * Correct positioning of a parse area reset to ensure all assignment calls perform the reset function. * Correct an old bug related to IMACRO support. FileWatch ReLoad and MEdit were not honoring a IMACRO xxxx OFF setting. Latest EXE version Download
SPFLite3.3.1.24113.exe (508 KB) I hope some of you can find the time to participate. George
|
|
|
Post by George on Apr 8, 2024 13:02:31 GMT -5
Robert: I had telemetry code installed and working in test versions, but both you and Stefan convinced me to remove it over privacy concerns, since the code had to transmmit data back 'home'.
So we're stuck with begging for feedback.
George
|
|
|
Post by George on Apr 8, 2024 12:08:51 GMT -5
Hi, I've been reviewing the Doc. for an upcoming release, and have been browsing the whole file, looking for any impacted sections.
I've come across one feature we put in quite a while ago, and I have serious doubts that it is ever used, the "Working with" section describing it is 8+ pages of pretty dense text.
So ... does anyone actually use "Command Chaining" (and Command Chaining does not mean using a ; to separate individual commands, that's totally different).
I wrote the code to support it, and other than testing it, have never used it, or even thought about using it. Being honest, it would simply not come to mind since I'd have to read the whole Help section before attempting it.
I'd really appreciate comments on this. The release I'm working on is a big cleanup release, I'd like to kill this.
George
|
|
|
Post by George on Apr 7, 2024 13:22:26 GMT -5
Robert: I'm close to finally putting out a Beta (No -- an Alpha). I'm going to make it a whole major jump to SPFLite3 and a separate install folder, so two versions can co-exist. This has been a really interesting exercise. Amazing how many unreported bugs I've uncovered.
To prevent any possible corruption of the CFG file, the new version will (for a while) make a timestamped backup of the CFG file at each invocation.
Also, your suggestion for the AA.Macname / BB.Macname is included. Any prefix will override the normal Profile MACLIB setting.
George
|
|
|
Post by George on Mar 31, 2024 8:29:44 GMT -5
MUEH: Not surprised. Message handling received some significant changes to correct other problems. I'll have a look to see how to fix it.
George
[UPDATE]
Corrected. The previous change delayed clearing any error message until some user action (KB/Mouse) took place in the tab. In your case, the post macro DO commands included a (Home) key - which - cleared the message. On my system you can just see the message blink on the screen before being erased.
I've altered it so the clearing will not occur if a set of DO commands are executing.
I'll post a fixup Beta.
George
[/UPDATE]
|
|
|
Post by George on Mar 31, 2024 8:27:22 GMT -5
Robert: That's why I only added UTF8. The code for UTF16 is in there, but it's embedded in the main edit file read routine, so not easily accessable. I didn't think it worth the effort to clone it in another spot. Besides the justification just isn't there.
George
|
|
|
Post by George on Mar 30, 2024 11:24:58 GMT -5
A Beta with the correction will be uploaded today to correct this. NOTE: there will be no support in INCLUDE for UTF16/32 format encoding.
George
|
|
|
Post by George on Mar 30, 2024 10:59:02 GMT -5
Well, it's almost certainly BOM data. The #INCLUDE file reading simply ignores the possibility of BOM. I'll have to add it.
I'll do so and play with the provided data files. The Pli file is definitely has a UTF8 BOM marker.
George
|
|