Housekeeping

tahanson43206 · 2021-07-01 18:50:41

For SpaceNut re #1325

OK ... The existing procedure bans the account ....

It would mean a bit extra work for you, but the safety feature is there, so we can use it as is.

Would you like for me to prepare a list of usernames for your review?

I'll wait until after the current set is finished. That will happen Saturday evening, if all goes well.

The collection will be on the order of 7,000 records, based on my preliminary investigation.

I can pass the list to you via NewMarsMember portal, or via DropBox if you prefer.

Edit#1: The present run will end (if all goes well) at an odd number, of 81 in the last group.

Would you be interested in a test run of 19 from the pool of Zero Post users?

Per your request, the TestID's would end up in the Bans table, ready for your review.

That would save you having to ban them in the first place, ** and ** it would insure the protection that banning provides.

(th)

Last edited by tahanson43206 (2021-07-01 19:07:50)

tahanson43206 · 2021-07-02 08:54:23

For SpaceNut re ID Recovery ...

The Mouse Trainer is undergoing enhancement ...

The specific feature in focus is Ratio Discovery

The first test confirmed the value of the safety limit, because an infinite loop occurred.

The safety limit worked as intended.

Inspection this morning revealed that while mouse positioning commands were computed and sent, the Mouse Click command was not issued.

That oversight is corrected, so the next live test will reveal any other defects that may exist.

Updates at 15:38 local time

FluxBB.org appears to be back up with Certificates... That makes sense (to me for sure)

Mouse Trainer is proving to be a bit of a slog ... Today I took it to a machine where I could debug with the system debugger, and discovered/learned that a system function to convert strings to numbers will balk if given a null string. That was easy enough to correct for.

However, the trial positions are still showing up in the wrong locations (ie, outside the Reference Frame) so I'll do some touchup tomorrow to fix that.

The coordinates of the Reference Frame are available, so it should be possible to use that information to bring the AACKeys trials back inside the Frame.

Tonight's run to 10700 should queue up around 22:00 local time.

Update at 22:23

The Internet is really congested this evening! The smart wait feature of WBA can tolerate some delay, but it can't deal with lost packets.

The program managed to complete two ID's by itself, and I had to help with two.

I'll try again tomorrow. We only have 177 usernames to go. A delay of a few hours won't be a problem.

Then we'll have an opportunity to look at the next 7,000 or so candidates for upgrade.

***
The Mouse Trainer problem of trying to match up two coordinate systems is proving a bit more challenging that I would have expected.

Tomorrow I'll try to put a fence around the guessing process. The Reference Frame has borders (Top, Bottom, Left, Right).

When I send a command to AACKeys it can put the mouse inside or outside the Reference Frame, and (it turns out) AACKeys will be perfectly happy to accept coordinates that are outside the physical system itself.

The feedback mechanism is able to tell me if the guess is inside the Reference Frame, by returning integers greater than zero.

If the guess sends the mouse outside the reference frame, nothing is returned.

The Reference Frame is small compared to the Universe of numbers AACKeys will apparently accept.

My thinking for tomorrow is to find the boundaries of the Reference Frame before attempting to find a point inside the frame.

(th)

Last edited by tahanson43206 (2021-07-02 20:35:34)

tahanson43206 · 2021-07-03 09:40:34

For SpaceNut re ID Recovery ...

The Internet had calm waters this morning ...

WBA completed the run to 10,700 without further incident.

The Smart Wait feature could be enhanced to better manage choppy Internet conditions, but it would be a lot of work to achieve that when simply waiting a few hours is a valid solution.

Completed Sequence for ID: 10700
Total Command Lines found: 111
Total input Lines in script: 337
Total input Lines in data: 178
Number of ID's processed: 96
Starting Number: 10605
Last Number of Run: 10700
Summary for Web Automation Report for 07-03-2021 at 11:31:32
Average time of Loop from Main form: 00:01:27
Total time of Processing: 02:20:14
Total time Program was Active: 02:31:52
EOJ Msg#1: Successful Run. No Warning Errors.

***
Mouse Trainer is coming along ... code to find the borders of the Reference Frame using System Coordinates is working. It will be a bit more challenging to find the corresponding coordinates in the AACKeys generic coordinate system, but I think the procedure will be LESS challenging than the current attempt to find a point.

Knowing the boarders of the generic window will help the point finder by revealing invalid guesses before they are attempted.

***
This is the weekend in NewMars forum land ,... I'm looking forward to member contributions by those who have a bit more free time.

***
Just 81 names remain in the Username input file. I'm tempted to polish them off now, since conditions appear to be favorable.

(th)

Last edited by tahanson43206 (2021-07-03 09:43:01)

SpaceNut · 2021-07-03 09:49:03

I see you are working from the back of the list z descending of the banned list towards Thomas....I read 4 plus pages of 50 before getting to Thomas. and about a page after that until we get to the testid's

NewMarsMember · 2021-07-03 10:06:33

For SpaceNut re #1329

Thanks for your notes on the progress of the current Username list ...

The sort order is more or less accidental ... I ** think ** this particular batch is in ascending order ...

The final set of 81 Usernames is in process. I am looking forward to finishing work on the Bans and starting on the Zero Posts group.

I'll collect the names of candidates first, and then inquire how you'd like to proceed.

It would (???) make sense to try a short run of just 19, to see how it goes.

Per your request, I would use the existing script which Bans the users after converting them.

That would provide the safety factor you asked for.

(th)

Last edited by NewMarsMember (2021-07-03 10:07:05)

tahanson43206 · 2021-07-03 11:52:15

For SpaceNut re ID Recovery ...

The run to 10781 finished with a hiccup on the last username ... It contained a space, which stopped the run.

I finished the last username manually, so we now have completed a complete pass through the Bans table, selecting for ID's with Zero Posts.

I'll next set up a read-only script to pull the usernames with Zero Posts.

Starting Number: 10701
Last Number of Run: 10781
Summary for Web Automation Report for 07-03-2021 at 13:48:12
Average time of Loop from Main form: 00:01:27
2 Date Exceptions were recorded.
Total time of Processing: 01:57:54
Total time Program was Active: 02:03:47
Early end of run requested.

(th)

SpaceNut · 2021-07-03 12:27:35

Starting in the B's for those banned and zero posts is username BrokersLaups.

Next group is in the D's for zero posts.

edit

have been adding note: "Account locked" with the ban of single posts users

was able to process 35 pages of names setting the low counts to zero if they were accounts to convert

tahanson43206 · 2021-07-03 12:59:48

For SpaceNut re ID Recovery

This is your project, so I'll work with Bans if you prefer.

The advantage that ** I ** see is that you are taking the time to examine each candidate before making the decision to ban them.

While this process will take longer, it is ** definitely ** safer than letting a program run through the candidates.

The script to collect user names for the next round is probably done, so I'll take a look to see how many candidates it comes up with.

There ** is ** another option .... the program can ban all those candidates without upgrading them to TestID format.

That way, you can unban the ones you want to keep, and they will be immediately removed from "dangerous" status.

If you manually ban all 7,000, it will take a while, during which the spammers could log in unimpeded.

Again, your show.... just let me know which group of records you want me to work with.

Update at 16:31 local time

Completed Sequence for ID: 361
Total Command Lines found: 12
Total input Lines in script: 56
Number of ID's processed: 361
Starting Number: 1
Last Number of Run: 361
Summary for Web Automation Report for 07-03-2021 at 16:28:38
Average time of Loop from Main form: 00:00:21
Total time of Processing: 02:06:40
Total time Program was Active: 02:28:39
EOJ Msg#1: Successful Run. No Warning Errors.

The run above collected all Usernames in 361 pages delivered by the Zero Posts query.

I'll remove TestID's to see what the grand total looks like

Update at 16:43

The scan for Zero Posts came up with 11,750 after TestID's were removed.

There should have been just 7,000 or so. There must be duplicates.

I'll run a procedure to remove any duplicates to see what might be left.

Update at 18:34 local time ...

The file of 11,750 usernames that is ** supposed ** to represent users with Zero Posts contains plenty of duplicates.

The procedure that collected the usernames seems simple enough ... just collect the names 50 at a time as they are served by the query.

The wrinkle (in this case) is that the query is re-run (supposedly once) for each page ...

I'll run a utility to remove the duplicates. I'll also take a look at the results from the run, to see if I can understand where the duplicates are coming from.

It may be that the strategy of asking for successive pages is causing FluxBB to become confused. It may need more time (more seconds) between calls.

(th)

Last edited by tahanson43206 (2021-07-03 16:40:02)

tahanson43206 · 2021-07-04 06:47:44

For SpaceNut re ID Recovery ...

This post is reserved for updates on updates ...

The Generic Utility will be updated to select unique rows from sorted input. This capability is reported to be present in modern spreadsheet packages, but in a quick search of packages immediately available, I didn't find it. It is advertised as being an option under Data >> Remove Duplicates.

The Mouse Trainer is on the verge of a breakthrough in solving the problem of matching a generic virtual coordinate system (AACKeys) to the real world physical coordinate system of a real computer. The problem is easily solved if there is a human in the loop, but I'm trying to enable to program to solve the problem ** without ** assistance. The problem from the computer's point of view is that there is no easy way for the computer to know where the virtual coordinate system is at the beginning of the operation. Whereas a human can see (optically) exactly where things are, the computer is going to have to find the Reference Frame on the physical layout by guessing where it might be.

The Reference Frame might not even exist on the physical structure. It is possible for an operater to place the reference frame outside the physical structure.

My tentative solution is to perform a diagonal search from Top-Left to Bottom-Right of the virtual space, and hope to hit the reference frame.

If there is no hit, then the attempt will end and the operator will be given an opportunity to intervene.

If there ** is ** a hit, then it should be straightforward from there to find the borders of the Reference Frame, because feedback is only provided when the mouse clicks are inside the reference frame. Meanwhile, the physical structure is (hopefully) holding steady, so the set of points that connect the two systems can be established fairly quickly.

The goal of the exercise is to find the ratio in X and the ratio in Y between the two systems. Those numbers can then be used to quickly and accurately create script commands as needed for a particular situation.

Update at 14:37 local time ....

In working on the output of the recent lookup only script to find users with zero posts, I find that only 4700 names survive after duplicates are removed.

There should have been 7,456 based upon the FluxBB query. I've decided to redesign the query ... the initial simple method obviously overwhelmed the server, leading to thousands of duplicates and thousands missing.

Nice to see discussion of the Flying Car and building design!

Noah sent me references to go with a paper he wrote that he invited me to review. I need to set aside a few hours to do justice to his efforts.

My impression is that he is energetic and hard working but has not yet received guidance from anyone on how to write a paper.

There may be books available on how to write an academic paper. My impression from seeing peer reviewed papers is that there must be a routine that authors are trained to follow that leads to success, but I never received that kind of coaching. It sure would be helpful to know if there is a publicly available resource that anyone writing a paper can follow.

Failure to follow whatever the guidelines for Best Practice might be leads to an unfortunate similarity to plagiarism.

As a rule of thumb, it is my impression that any text pulled from a resource must be clearly marked and referenced both at the time and in the appendix.

More than that though, is the need to show references for statements about the nature of the Universe that must have been discovered by scientific method and documented thoroughly somewhere. An author can make assertions about how the Universe works, but without references to back up such statements, the words ring hollow.

An exception might be to report something that is generally accepted as true, such as that the sky on Earth looks blue to certain persons when certain conditions exist. Even there, a scientific reference would not be out of order, although I admit it might be regarded as unnecessary.

(th)

Last edited by tahanson43206 (2021-07-04 12:49:20)

SpaceNut · 2021-07-04 14:57:39

The duplicate is on the email is that correct?

have cleared and banned through to page 42 into the H's

have verified all low counts and reset those with no posts to zero

have flag those with low counts as locked as true poster that need outreach later.

tahanson43206 · 2021-07-04 16:52:47

For SpaceNut re #1335

Thanks for your report and for the question ...

The duplicates problem I was seeing was caused by running the script too fast. i just reran it with 30 seconds per page of usernames, and the results look much better.

However, your report of progress manually banning spammers means that I can run a script to lookup candidates in the Bans table through the last letter you've finished ...

I can run a script on the bans table through the H's easily enough ... i'll tackle that tomorrow.

Thanks for taking a look at the ID's with posts showing but no actual posts.

At the rate you're going, I should have at least a few days work in the queue.

***
It's good to see Calliban's interest in the floating city concept.

Such a concept only makes sense if the structure has nuclear power as a primary source.

Solar power might be input to the system where it is exposed to the sky, but most of the system will be underwater like an iceberg.

***
I'd like to encourage you to consider building a small scale model of the water evaporation machine that you showed us from the MIT labs. You should be able to feed your well water into it and receive pure water out.

The amount might be small, but it would be an encouragement.

***
New Hampshire made the national news recently .... trying to remember what it was about ... New Hampshire used to be in the news frequently because of a famous Senator who kept getting re-elected.

For a small state, New Hampshire sure does seem to operate at a heavy weight level much of the time.

Update at 19:03 local time .... the run with more time (30 seconds) between fetches gave better results.

The list of usernames with Zero posts is 7248, which is pretty close to the number that FluxBB quoted.

The way I might be able to use this list is by revising the main script to make sure a username is in the Bans table before working on it.

It would (probably) be easier to do that than to fuss with the Bans table itself, which is cluttered up with so much debris from 20 years of spammer activity.

I'll hold off until I can study what is needed, and then confirm with you before proceeding.

Per your request, the work would only be performed on records that are in the Bans table.

Roughly (thinking out loud here) ...

1) Read a name from the input file
2) Check that it is present in the User table
3) Check that is is present in the Bans table
4) Go back to the user table and perform upgrade there
5) Return to the Bans table and complete finishing steps there

That script may be quite close to the main update script we've been using.

Still thinking out loud ... a new script might make sense ... it would perform the first three steps and report any exceptions. That way, the script we have could run without change.

I'll pursue that. it would be safer.

Update at 21:49 local time ... the search for the Reference form along a diagonal from 1,1 worked as hoped.

The mouse moved across the left monitor in steps of 100 across and 10 down until it landed in the Reference Frame and immediately halted with a success message.

Tomorrow I'll add in the code to find the borders of the Reference Frame.

(th)

Last edited by tahanson43206 (2021-07-04 19:51:01)

tahanson43206 · 2021-07-05 08:08:45

For SpaceNut re ID Recovery ...

I have two mini-objectives in mind for today ...

The quickest to complete would be a test of a new script to confirm that usernames I've just collected (those without posts) are in the Bans table.

The situation may need a new command, and therefore a modification of WBA ...

At present, the command available looks for text in the clipboard, and reports if it is present.

What is needed is the reverse ... I want to detect any usernames NOT in the Bans table.

On the other hand, the existing procedure will deliver a list of usernames that ARE in the Bans table, and that can be massaged to deliver a work list for updates.

So I've just convinced myself to try that.

***
The ** other ** activity planned for today is to build on yesterday's success finding the Reference Frame by using a diagonal search from the upper left of the physical layout. If the operator understands the process, then they can make sure the Reference Frame is in the path of the search, and Bingo! will occur every time.

The procedure for today is to find the borders of the reference frame to get a set of preliminary values for X and Y ratio between the AACKeys generic coordinate system and the physical structure where it is running. With those it should be easy to find precise values for the ratios, suitable for calculation of coordinates for script commands.

Update at 12:40 local time ...

For SpaceNut just FYI .... WBA is running a new script to find usernames in the Bans table.

The input to the run is a list of all users with Zero posts (omitting TestID's) and the output will be users with zero posts you have banned.

Running this way avoids the debris in the Bans table.

If a username is in the users table (prerequisite) AND if it is also in the Bans table (requirement) then I can update it.

The first run is for 960 users names from input of 7,456 candidates.

The run is working from the low end of the alphabet, so it will pick up your work in the D's-H's.

Inside Baseball: This script runs as a Guest. Login is not required to see "Banned" status.

***
This is housekeeping so I'll toss in a 4th of July related item ...

For years I've grilled on the 4th in the traditional way, just before meal time.

Today, I'm adapting slow cooking that has been working well here in recent months.

I just completed the grill, and put the result into the slow cooker. It can sit there for hours (just above a water bath).

I have no idea how it will turn out, but am guardedly hopeful.

Update at 19:22

The test run of a script to find Banned users in the Bans Table using Zero Posts users as input ran for 8 hours and turned up nothing. I had misspelled "Banned" as "banned" ... picky, picky!

A short run of 10 usernames with the correct spelling was more successful - 10 records for 10 tested.

Fortunately (to the best of my knowledge) there was no cost other than a bit of electricity consumed at various locations in the long chain between here and the server.

I'll relaunch the script with a larger search number.

It will run quietly overnight, if all goes well. Each lookup takes 24 seconds, of which most is wait time.

(th)

Last edited by tahanson43206 (2021-07-05 17:26:27)

SpaceNut · 2021-07-05 19:21:01

Just had a visit from patriothmw one of those that when checked via google is a russian spammer...

Time to lock the doors on more of the same....

tahanson43206 · 2021-07-05 19:56:24

For SpaceNut re #1338

Thanks for catching patriothmw live! I can't help thinking the gent behind that spam id was amusing himself by launching it on our national holiday.

We have 7000+ zero posts to go ... If you move 100 a day to the Bans table, WBA should be able to keep up.

I expect to resume upgrades tomorrow evening, assuming the verification run succeeds, now that I spelled "Banned" correctly.

***
Tomorrow's a work day for you and your co-workers, so I hope the short week goes well.

I saw a report (headline) indicating that Iceland has been experimenting with a four day work week and they are (apparently) happy with the results.

With much of the labor of modern society now being carried out by computers, I can understand that it might be possible to cut back on human work time.

The fact that people have another day of free time doesn't mean they aren't productive. Instead, they are able to spend more time doing things ** they ** want to do, and often those activities benefit society one way or another.

(th)

SpaceNut · 2021-07-05 20:24:14

Did some more clearing house of id's and found the invitedguestspeaker1 id; which I made sure not to ban or indicate locked for low posting....

Early when I took the job where I am it had what was a 5 day one week followed by a 4 day week the next that were 9hr days.
I liked it so that we would be able to schedule medical appointments and enjoy a long week end.

Speaking of medical care the local hospital had satelite doctor offices and after its sale the many offices are have trouble keeping staff and a few of them have closed forcing patients to go to the ER instead that will make them greater profits for the new owners.

Its bad enough that medical care became band aids so no wonder we have been hit so hard with covid....

tahanson43206 · 2021-07-06 06:29:54

Thanks for saving the Invited Guest ID ... It seems to me that is still a good idea, if we can find the right combination of speaker and forum members who want to engage with the speaker.

Today's verify run ended a few minutes ago ... the program ran for over 12 hours, and processed 1800 usernames.

I'll find out how many of those are eligible for Upgrade shortly. That's 18 days of work for WBA if all were both in the Users table AND in the Bans table.

Update for Web Automation Report for 07-05-2021 at 19:28:59
WebBrowserAutomation
Version 2021.6.9.1
1800 count Verification Script run ...
Query Text Requested was Not Found: Hermanvah
Last username tested: Hermanvah
Completed Sequence for ID: 1800
Total Command Lines found: 26
Total input Lines in script: 76
Total input Lines in data: 7248
Number of ID's processed: 1800
Starting Number: 1
Last Number of Run: 1800
Summary for Web Automation Report for 07-06-2021 at 08:19:56
Average time of Loop from Main form: 00:00:25
Total time of Processing: 12:18:23
Total time Program was Active: 12:50:13

No hiccups but this was a relatively simple script.

I did see the Smart Wait working every now and then ... that feature allows for occasional delays in response from FluxBB. What that feature cannot (currently) handle is a dropped packet, which I reported seeing a few days ago.

Edit: The standard Internet has two modes of operation. A broadcast mode does not have provision for packet delivery verification, and that is suitable for videos, for example. The other is the more common mode, which provides for verification that each packet transmitted was received. Packets can arrive out of order, so they have sequence numbers. However, if two packets are dropped, then the connection is dead.

The sender sends a packet and expects a confirmation of safe arrival, but if that packet is lost, and the request for retransmission from the receiver is lost, then we are done. I'd have to go read the specs to be sure, but I assume there is provision in the protocol for retrying after a packet is lost, but there has to be a practical limit to that, and we saw that during a heavy congestion episode a few days ago.

Inside Baseball: The verify run had 1800 usernames and 1797 were reported as being in the Bans table.

I'll start the WBA runs again with the 1797 as input.

I'll schedule a short run of 19 today, to bring the TestID total back up to an even 100.

Update at 9:27 local time ... the short run to 10800 appears to have completed without incident.

I'll schedule a run to 10900 for around 22:00 local time this evening.

***
Testing revisions to the Mouse Trainer is next on the local agenda ... Diagonal search for the Reference Frame is working.

The current objective is to find the borders of the Reference Frame.

Completed Sequence for ID: 10800
Total Command Lines found: 111
Total input Lines in script: 337
Total input Lines in data: 1797
Number of ID's processed: 19
Starting Number: 10782
Last Number of Run: 10800
Summary for Web Automation Report for 07-06-2021 at 09:36:46
Average time of Loop from Main form: 00:01:25
1 Date Exceptions were recorded.
Total time of Processing: 00:28:35
Total time Program was Active: 00:46:35

(th)

Last edited by tahanson43206 (2021-07-06 07:39:02)

tahanson43206 · 2021-07-06 13:22:06

For SpaceNut re ID Recovery ...

This post is specifically about what I am learning as I attempt to improve the performance of the Mouse Trainer, but I'll keep it in reserve for a report on tonight's WBA run to 10900.

Opening ... set stage:

The English measuring system and the Metric measuring system are relatively stable.

The ratio between an English inch and Metric centimeter is 1:2.54 (and change)

That ratio holds regardless of the lengths compared, so long as the distances are the same.

For example, 1 inch, 6 inches or 12 inches all have a 2.54 ratio relationship to their corresponding Metric measurements.

Going into the Mouse Trainer exercise, I've been dealing with the difference between a virtual reference system used by AACKeys, and the physical coordinates of a particular physical computer.

I was (surprised/dismayed) to discover that the ratio between the physical coordinate system and the physical one does ** not ** appear to be constant.

It will take some digging to find out what's going on. I assume (at a minimum) that the number of physical pixels on a monitor is a factor.

In addition, it is possible for the operator to configure display software to display at density different from the physical capability of the monitor.

I am running two monitors on the Dell in use for production. I've not had a reason to pay attention to either the physical capability or the software settings, other than to make sure the displays are readable.

In addition to the above, I appear to have discovered that an AACKeys virtual "screen" extends well past the physical one.

I'm currently working on finding the borders of the Reference Frame (which is laid over the Dell monitor), and to my amazement, the ACCKeys mouse position algorithm went happily past the end of the physical space for a considerable distance, before finally deciding it had reached the end.

The one saving grace, and the anchor that I'm holding onto, is that regardless of the underlying ratio variation, once a particular point is matched between the physical environment and the virtual layout, the ratio holds steady. It is ** that ** feature that allows WBA to consistently hit fields and buttons on the NewMars FluxBB screens as needed for a particular script.

The points were obtained using a manual guessing process. The software I'm trying to develop ** may ** be able to accomplish that, but at the moment, the squishy nature of the environment is proving challenging.

(th)

SpaceNut · 2021-07-06 20:58:08

have cleared id's 8101 through to 8200

seems only one id was missed in the b's with a zero post count that was banned so the program did good enough as we can clean up what is missed later.

Bruce 3 Rossman

tahanson43206 · 2021-07-06 21:21:13

For SpaceNut re #1343

Thanks for showing me the missed username ... I notice two spaces in the name ...

The program is able to handle one space ... I wonder if having two caused the miss.

***
Congratulations on polishing the ID's through 8200!

WBA is staying just a bit out in front, with 10,900 this evening.

Completed Sequence for ID: 10900
Total Command Lines found: 111
Total input Lines in script: 337
Total input Lines in data: 1778
Number of ID's processed: 100
Starting Number: 10801
Last Number of Run: 10900
Summary for Web Automation Report for 07-06-2021 at 23:16:01
Average time of Loop from Main form: 00:01:29
17 Date Exceptions were recorded.
Total time of Processing: 02:29:12
Total time Program was Active: 02:31:46

(th)

tahanson43206 · 2021-07-07 06:30:30

For SpaceNut re ID Recovery ...

This post is reserved for updates on Mouse Trainer .... The problem of trying to match the virtual coordinate system used by AACKeys to the physical coordinate system of a particular combination of hardware and software has turned out to reveal complexity of which I have been blissfully unaware.

It appears possible that there are "forces" (in the sense of characteristics of hardware and software) at work to cause the ratio of X to X (and Y to Y) in the two systems (overlaid on each other) to vary. It would be nice to discover that whatever the forces are they can be discovered and used with confidence.

Louis made a memorable comment about forces in a post, long ago and far away in the NewMars archive. His (perceived by me) plea for assistance in understanding the meaning of the concept of "force" has stayed with me. It may turn out that the ability to understand force is genetic. It may turn out that without the right neuron structures, understanding may be impossible.

In the current situation, I am using the word "force" in the sense of influence ... ** something ** is influencing the behavior of the mouse positioning software, and at present, I do not know what it is.

My ** theory ** at the outset was that the virtual coordinate system of AACKeys is static ... the distance between two successive points in X or Y is the same over the entire field. I am no longer sure that is the case.

Likewise, at the outset of this investigation, I had (probably naively) assumed that the physical layout would remain static and uniform in dimension throughout its range. However, changing pixel counts in hardware, and changing pixel aggregations by graphics software, may yield variation as perceived by AACKeys.

Results coming in from experiment are suggesting the starting assumptions were not correct.

Today, I'll add code to record the ratio between the coordinate systems at multiple points (every 10 to be exact) in both X and Y. If I cannot deduce the nature of whatever "force" is at work, I can at least used the ratio at a particular point by pulling it from a table.

That is basically how traditional celestial navigation works ... the ephemeris tables are good for particular moments in time.

Such a table would be reliable for a particular combination of hardware and software. It would have to be regenerated when the application is moved to a different system, or if the one on which it is running changes.

Update at 11:49 local time ....

Attempts to keep the AACKeys mouse inside the designated playground remain unsuccessful, despite using four different techniques to try to detect the mouse breaking through the fence.

However, a preliminary result is available for what the Ratios of the two coordinate systems (may) look like ...

Update for Web Automation Report for 07-07-2021 at 11:22:28
MousePositionTrainingUtility
Version 2021.7.7.1
This was the first test run of 7.7.1
Reference Frame: Left border: 1021 These are system coordinates
Reference Frame: Right border: 2051 Difference is 1030 (Dell screen)
Reference Frame: Top border: -3 Not sure why negative here
Reference Frame: Bottom border: 771 Difference is 774 (Dell screen)
X Target for Ratio Procedure: 1467 Gmail button on Browser display
Y Target for Ratio Procedure: 327 Physical coordinates
Reference Frame Found by AACKeys X is: 501 Y is: 51
Ratio computations: System:AACKeys X: 1 Y: 1 <<= need format decimal
Left Border Found! AACKeys X: 381 X is: 381 Y is: 51
Right Border Found! AACKeys X: 981 <<== diff 600 - went past border
Computed midpoint of Reference Frame X: 681 (300 is half of 600)
Top Border Found! AACKeys Y: 9 <<== stopped by itself
Bottom Border Found! AACKeys Y: 609 <<== continued past border
todo:
1) fix ratio computation and display
2) add ratio computation and display to search
3) add logging to search along with ratio
Note hand computations:
Given Y display during X hunt: 135 System 51 AACKeys >> 2.647 ratio
Given X display during Y hunt: 1813 System 681 AACKeys >> 2.662

I was surprised to see the similarity for ratios for X and Y ... Aspect Ratio is determined by physical design of monitors.

If the AACKeys aspect ratio matches the physical one, I would **think** it would be pure coincidence.

On the ** other ** hand, the AACKeys program designer may have fetched properties of the system it's running on.

The only way (that I can think of) to test that is to test the program on another hardware configuration.

Edit/update at 16:30 local time ...

Mouse Trainer is giving reasonable results ... The end-of-search-loop procedures needed some tweaking ... In the end a comparison of the trial location with the known borders of the Reference Frame appears to have worked. It helped to feed the correct variables into the comparison.

Tomorrow I'll try to figure out how to handle the X:X and Y:Y ratios ... they definitely vary from one reading to the next, but the range of variation may be within boundaries that I might be able to work with.

The whole point of this particular exercise is to try to save time when preparing a script for a particular task. While computing the ratios is time consuming, once they are obtained (and saved) they should save time when actually preparing a script.

***
Tonight's run to 11000 should kick off at about 22:00 local time.

Update at 19:57 local time .... After taking 68 samples for Ratio computation, discarding one as outside of the norm, I found that the average for both X and Y is 2.65 and change .... The variation does range to 2..64+ and 2.66+, but the average looks usable.

I'll experiment tomorrow with a trial run using the computed (average) Ratios. Since the purpose of the exercise is to put the "Homing Procedure" into the vicinity of the target, I'm hoping that X:2.65 and Y:2.65 will accomplish that.

(th)

Last edited by tahanson43206 (2021-07-07 18:01:40)

tahanson43206 · 2021-07-07 18:59:39

For SpaceNut .... FluxBB.org announced an update to FluxBB

There are a number of items, but I thought you might like this one:

Prevent duplicate bans

We've seen a lot of those !!!

(th)

SpaceNut · 2021-07-07 20:29:48

That would seem to be a good upgrade since we are seeing that double email when changing the username....

have did the unban from 8201 through to 8350 tonight.
eyes are heavy and its time to go to sleep...

tahanson43206 · 2021-07-07 20:56:22

For SpaceNut re #1347

Thanks for putting in the extra time this evening!

Just FYI ... something curious happened...

Darrennog came up as missing during this evening's run.

That should not have happened, because the starting point for the current series was the users table.

I collected all users with zero posts as the first step.

That is where Darrennog would have come from.

Then I ran a verify procedure to make sure the names were in the Bans table.

Three names were NOT in the Bans table, so they were not passed forward.

Somewhere along the line between the first collection of names and now, Darrennog became non-available.

I'll start the run over without that name.

Update at 23:09 ... my input file is repeating itself....

I'll stop for tonight and straighten things out tomorrow.

Update at 9:52 local time ....

There are several factors that (I suspect) led to the halt last night. That input file should have been clean as a whistle, but it obviously was not.

The first defect in the procedure that jumps out is the lack of sequence checking in the duplicates removal feature I added to a string utility called Extract.

That is easy to fix, and I'll do that today.

Beyond that though, is the question of how a duplicate set of usernames got into the file in the first place. I suspect that I've allowed insufficient time between the command to highlight the area to be captured, the capture command, and the command to save what is captured to disk.

What is more ... I am NOT doing validation at all, for this (relatively simple) lookup script. That is something that can be corrected, so I'm planning to do ** that ** the next time I run that particular gather script.

My guess is that the failure involves repetition of precisely 50 usernames, which would be one query page.

That guess is based upon the sum of 19 and 30 or 49 ... that is the sum of the two runs completed before the halt.

(th)

Last edited by tahanson43206 (2021-07-08 07:58:52)

tahanson43206 · 2021-07-08 11:29:25

For SpaceNut re ID Recovery ...

1598 user names were left after scrubbing the input file that failed due to duplicates ...

The defect in the duplicates detector utility is on the work list for correction.

A run to 11000 is in progress. Hopefully it will complete without further incident.

An input file of 5,000+ user names will be scrubbed as well, because the failing file was pulled from that one.

The script that generated the duplicates in the first place is on the work list for correction.

***
Update at 15:09 local time .... the run to 11,000 completed ...

Text direct from Clipboard in Step: 34 DemetriusL
At Step: 42 Username did not match Stop At value
Post-Date Function Start: Reading Clipboard #4
PerformFunction: Valid ID: match in:
Registered: 2017-06-25 (107.150.91.72)
Last post: Never
Last visit: 2017-06-25 03:12:40
Posts: 0 - Show all
Text direct from Clipboard in Step: 54 TestID11000
Text direct from Clipboard in Step: 63 TestID11000@test.com
Completed Sequence for ID: 11000

Total Command Lines found: 111
Total input Lines in script: 337
Total input Lines in data: 1598

Number of ID's processed: 70
Starting Number: 10931
Last Number of Run: 11000
Summary for Web Automation Report for 07-08-2021 at 15:07:46
Average time of Loop from Main form: 00:01:28
6 Date Exceptions were recorded.
Total time of Processing: 01:44:46
Total time Program was Active: 01:44:51

Hopefully the input file is clear for the rest of the immediate set

(th)

Last edited by tahanson43206 (2021-07-09 05:38:39)

tahanson43206 · 2021-07-09 05:40:30

For SpaceNut re ID Recovery ...

The WBA had a hiccup on the last username of last night's run, so I decided to reach the "all one's" goal

Completed Sequence for ID: 11111
Total Command Lines found: 111
Total input Lines in script: 337
Total input Lines in data: 1428
Number of ID's processed: 12
Starting Number: 11100
Last Number of Run: 11111
Summary for Web Automation Report for 07-09-2021 at 07:36:57
Average time of Loop from Main form: 00:01:23
2 Date Exceptions were recorded.
Total time of Processing: 00:18:12
Total time Program was Active: 00:20:14

Tonight's run should put us back on even 100's at 11200.

Update at 11:53 local time .... Mouse Trainer went through a test sequence on the Dell this morning ..;.

Ratio Discovery is working reliably, and I'm now edging toward being able to use it to pin locations the operator wants to visit on the Target form.

Early results indicate the program needs a number of adjustments. For one thing, it said it worked but I have no idea where the mouse went.

(th)

Last edited by tahanson43206 (2021-07-09 09:56:17)

New Mars Forums

Announcement

#1326 2021-07-01 18:50:41

Re: Housekeeping

#1327 2021-07-02 08:54:23

Re: Housekeeping

#1328 2021-07-03 09:40:34

Re: Housekeeping

#1329 2021-07-03 09:49:03

Re: Housekeeping

#1330 2021-07-03 10:06:33

Re: Housekeeping

#1331 2021-07-03 11:52:15

Re: Housekeeping

#1332 2021-07-03 12:27:35

Re: Housekeeping

#1333 2021-07-03 12:59:48

Re: Housekeeping

#1334 2021-07-04 06:47:44

Re: Housekeeping

#1335 2021-07-04 14:57:39

Re: Housekeeping

#1336 2021-07-04 16:52:47

Re: Housekeeping

#1337 2021-07-05 08:08:45

Re: Housekeeping

#1338 2021-07-05 19:21:01

Re: Housekeeping

#1339 2021-07-05 19:56:24

Re: Housekeeping

#1340 2021-07-05 20:24:14

Re: Housekeeping

#1341 2021-07-06 06:29:54

Re: Housekeeping

#1342 2021-07-06 13:22:06

Re: Housekeeping

#1343 2021-07-06 20:58:08

Re: Housekeeping

#1344 2021-07-06 21:21:13

Re: Housekeeping

#1345 2021-07-07 06:30:30

Re: Housekeeping

#1346 2021-07-07 18:59:39

Re: Housekeeping

#1347 2021-07-07 20:29:48

Re: Housekeeping

#1348 2021-07-07 20:56:22

Re: Housekeeping

#1349 2021-07-08 11:29:25

Re: Housekeeping

#1350 2021-07-09 05:40:30

Re: Housekeeping

Board footer