I've taken something of a large machette to the code and stripped everything to do with the group model leaving with just one set of user code for my role based memory model. There is a lot of information on how and why the role model was important, and that is all I have been testing since namespace was moved into bitweaver, so simply culling unused code made sense. The result is a lot more compact setup at least for the users package and just one style of memory handling to worry about. 20 years ago what I should have done is simply cloned the users module and built an alternative, and then fixed everything else by a simple switch between group and role ... hindsight is a wonderful thing! Having tidied everything up and mass hit all the php8.5 style issues using php-somthing I've now got to a point where I can start hitting some of the longstanding niggles in bitweaver's operations.

One long time niggle has been that {code} blocks were not working, and with Claude Code's help that was polished off in no time, and although it is still clunky, even ckeditor can cope with it. 

All of this overlies the heart problem that was diagnosed back in 2021 and which I am still waiting for a proper explanation of. The HIT session was intended to get my heart rate up since personally I think that is a problem, and while last year my heart rate had been dropping below 30BPM every night and the smart watch stopped recording, this year things seem a lot better and I've only had a very few days with 30 in the spreadsheet from the watch, which the finger probe backed up with 26 and 27's.

The latest session I managed to hit 149 during the first two 30 second sections of the hit, but then it dropped back to around the 100 mark so most of the rest of that section was 'zone 1' level. The exercises don't change, but the results vary hugely day on day. I am planning to have a proper discussion with my GP when she as the latest results in, although I am still waiting on an appointment with the hospital physio team to assess the current state of may knee. That and the cardiac consultant should be seeing me for an annual review. As always ... we just wait ...

We have logged a useful todo list
Completed:
✅ Second 8TB SAS disk installed smoothly  -> ASR-5405Z Configuration Notes
✅ cgit working on git.rdm1.uk -> Fun with cgit and cgi services in general
✅ fcgiwrap .socket -> .sock fix restoring lsces.uk cgi
✅ auth_basic working cleanly on subdomain
✅ webstack commits to tidy up -> Webstack - Server Configuration Management
Soon:
⬜ VSCode Remote-SSH setup
⬜ Development repos on srv10 -> sorted in the crib sheet above - have to keep everything on the one level, so the packages in /bitweaver get linked direct to /srv/git
⬜ cgitrc sections for repo grouping -> sorted in the crib sheet above
⬜ srv9 pull latest webstack commits -> which actually worked perfectly to bring srv9 in line with srv10
⬜ desktop webstack setup

Longer:
⬜ BitThemes APCu duplicate css/js bug
⬜ Smarty {code} / ckeditor fix
⬜ dhparams.pem consolidation -> Done on both machines ;)
⬜ acme.sh renewal niggles -> duff entry in config file, but next step is to drop the older style certs and just use the ecc ones.
⬜ mapserv on lsces.uk
⬜ Health dashboard in Firebird/PHP
⬜ MERG parts management
⬜ Historic map viewer (Leaflet)
⬜ wordpress_hacks.conf refinement

Just to add to the fun, I messed up and did a zypper dup on the main server before having triald it on the backup machine ... No warning we now have PHP8.5.5 after PHP8.4.20 and of cause nothing is included for firebird/interbase! So all the sites go down. The good news is that the whole point of the webstack exercise was to ensure that the the backup machine was in sync with the main server and EVERYTHING had been synced so that srv9 is running perfectly while srv10 is tits up, so hook up the the router and switch the two machine so 443 is served by srv9 and 8443 by srv10. So I can edit this blog happily and it's working fine.

arcconf is available from Microchip although the links provided by both Sonnet4.6 and duckduckgo search no longer exist. I eventually found it's now location, or rather locations as the more recent V4 version seemed not to work, while the V3 version gave a help menu however said there were no controllers. I did a cold reboot, and magically arcconf V3 reported a controller and showed the attached disk. Fairly quickly this morning we worked through the steps and got the drive attached finally using jbod mode to create /dev/sde. A little hold-up as I was waiting on yast/partitioner formatting the disk, only to realise that it was waiting for another 'next' before it ACTUALLY started the format. Next step is to tidy up and see if arcconf V4 will actually work now that the controller is actually 'ready'. ASR-5405Z Configuration Notes will document all the relevant bits from the process if only to remind me when I get around to a fourth disk in the machine, which I've just realised I can now use the last slot in the 4 slot cage as previously I only had 3 spare SATA cables. Something good comes out of the mess.

Another area that has been fun recentrly is the problems playing videos that have been recorded over 20+ years and now don't play properly ... avidemux Updates and Notes contains the short version, but I need to pull some of the test stuff into that crib sheet.

Back in the 1980's I was happily working with Borland BCB as my development framework, and all of the systems worked off that. We had developed a simple graphics card that allowed multiple monitors to be hooked up and used as for information displays. Four 625 line RGB feeds of each card, and a number of ISA cards could be driven of the one machine. Novel at the time since multiscreen did not appeared in windows until a little later. The quality of Microsoft's software was poor and some of the language used in the driver software should never have been allowed to be published, but we got by and displayed systems running 16 and 25 monitor stacks at many locations at the time. Of cause today multiscreen displays are standard as my current 11520 by 2160 desktop attests. While demonstrating a 4 screen setup to the Benefit Agency, a chance conversation opened up one major path, and the Caller Management System was created. This started life running on a single 286 based computer talking to an array of simple alphanumeric terminals. The Benefit Agency had not at that time even started to use computers, with everything being done on index cards and paper. Waiting times were logged in hours and the early systems at least kept track of just how long people were waiting. A fuller history can be found on the LSCES website, so lets keep to the software side of things. One thing that the BA system was initially not allowed to do was to identify callers, they were just a ticket number, but fairly quickly that bar was lifted as staff realised that information on previous visits could save a LOT of time, so the data management side came important. Being Borland based, Interbase was the obvious choice to handle this although the decision to end of life it in 1999 caused a lot of problems which initially resulted in the code being open sourced and Firebird came into existence. Many critical services had been using Interbase and when the accountants realised they had made a mistake they tried to close the door again ... and failed. Borland changed hands eventually being taken over by Embarcadero in 2007. I did keep up the annual fees to keep the development framework working, but by then the alternate path had already appeared.

At the end of the 90's, computers were starting to appear on the counters and interview rooms of the BA, so a request was made to replace the keypads with something on the computer, and making that browser based seemed an obvious move, so adding PHP into the mix happened. I was involved with the development of the early releases of PHP5 and actually went live using the beta release rather than PHP4. Since Firebird was not yet active, we still had to rely on Interbase, but that was supported by it's own PHP driver. This was also supported by a PHP library ADOdb which has been my base since day one. The original CMS web based code worked happily into the 2020's and the last systems were only taken out of service with the closure of face to face offices as a result of Covid.

There had been a development of a much more powerful version of CMS which offered a lot more flexibility, but in the end it never got rolled out live, however it was used on several websites such as my son's Cotswold Security support system. The original base was tikitwiki, but as with the BCB based systems, the core developers had rather blinkered views and adding functions was somewhat frowned on, so I became involved with a fork called tikipro which was based on a modular approach to functionality. It was renamed bitweaver following complaints from the tikiwiki project and I see that project has now adopted simply tiki. The latest build of bitweaver is what is running this site with Nginx, PHP8.4 and Firebird5 backing it up, and the forked code is fairly well documented on github. In addition to the ADOdb based stuff I have been forced to with Laravel's illuminate and the current driver for that is also on github. This was because while I struggled to keep phpgedview working with PHP8, the more active fork of that, webtrees, had dropped ADOdb and move to illuminate. I have the family tree back up and running finally, with everything stored in Firebird which is comfortable at least.

James had a falling out with his original employer before setting up Cotswold Security, and to tide him over he took over Rainbow Digital Media. The history of this is documented on that site and many of the customers were converted to bitweaver as soon as it because apparent that their sites were not as well protected as he had been lead to believe. I ended up stepping in to wipe a lot of inappropriate or fraudulent content that was sitting on sites that used an assortment of frameworks that had not been maintained at all. Initially the hackers were still accessing the sites which was also fun, but once the various back doors were eliminated, often simply by disabling on line editing of content, we got things back under control. The move to using bitweaver provided a secure way to restore clients edit facilities and in many cases, the existing style sheets could be mirrored as a bitweaver theme very quickly. Although a few clients did pay for redesigns as part of the process. Following James's death and with a number of live sites having been moved to their owners own hosting, only a few sites remain on the RDM banner.

That brings us almost up to date and where I am today. One thing that I have been trying to add to bitweaver for many years has been the ability to use pdf files like any other content. Search engines do index the text layers if available, and having spent much time scanning 50 odd years of Model Engineer magazines, being able to view an index of them on line, search for particular content and then display a selected result actually opening it on the right page with it's find function available has been a dream. Dolphin on my Linux desktop can search the content ... along with libreoffice content as well ... but then one has to manually find the relevent page. So what I have achieved in the last couple of weeks is a search that includes pdf text and and in browser viewer that then highlights the search result. All I have left to finish now is protecting access to material that will require a login. It is working fine on the development machine, but getting the production servers to mimic it has been a little problematic. But I will crack it! In the meantime the public domain material just needs processing to read the existing pdfs and update the text field in the database ...

https://www.embarcadero.com/products/cbuilder original development framework
https://www.embarcadero.com/products/interbase reason I am now on Firebird 
https://firebirdsql.org/ link to history perhaps
https://lsces.uk/wiki/a+potted+history ? add https://rainbowdigitalmedia.uk/wiki/History+of+Rainbow+Digital+Media
https://www.bitweaver.org/ although some of the functions such as search no longer work.
https://en.wikipedia.org/wiki/Tiki_Wiki_CMS_Groupware Which needs an update since the project is now simply Tiki
https://github.com/lsces?tab=repositories all my code is freely available on line
https://myhomecloud.uk/webtrees/index.php?route=%2Fwebtrees%2Ftree%2FCaineHumphries even my mods to get webtrees working with Firebird
https://medw.uk/search/index.php?highlight=Shumatech&content_type_guid=&search=go at last can search pdfs on-line!
https://medw.uk/fisheye/image/1189?highlight=Shumatech and display them with the search active

I say that everything is stable, but that is not quite accurate. What prompted the upgrade to the video distribution was a problem which originally seemed to be with the displayport output of the computer. It would go through bouts of going black and then coming back. Swapping over the 4k monitors, the problem seemed to stay with the displayport, although after replacing the cable to that with a 4k HDMI adapter and cable, the problem moved to the OTHER HDMI output ... It was that which then prompted the ripping out of the rest of the old cables and making sure everything was 4k rated. Which at first seemed to fix the problem, but then things started playing up again. The key here is that I'd been trying to stabilise the desktop by only using the new 'wayland' display service and drivers. Which has been documented already ( Application UI scaling and KDE desktop styling and usability problems ). A reset to the machine kicked me back onto X11 although I did not immediately realise that. The flashing monitor was no longer a problem but the layout was wrong. I went to fix that and then realised I'd been here before ... a quick check confirmed that yes we were back on X11. Log out and back in on wayland ... unstable monitor back. So for the last 24 hours I've been back with X11 and not a single flash. Only the aggro caused by switching to the windows computer when the Linux desktop moves everything around as the monitor is 'no longer connected'. Don't you just love progress!

Applying all of the notes I've gathered while messing around with the font sizes on the desktop I'm in a lot better place than I was a week ago. Was the agrro cause by creating a new user necessary, possibly not, but with the additional notes that have allowed me to port the extra stuff I actually use, it has been helpful and I would not rule out another new user at some point, but WHY can't I just hit reset and restore the desktop to a clean state hardware and software wise? I think that this perhaps has something to do with 'themes' and it may well be I need to spend time building my own, but that is not high on the todo list at the moment. The question is just what should I target next? I think I need to get back to the web services upgrade to PHP8.4 and get all of this infrastructure modernised ... but the PHP7.3 build is working fine, it's only a few bugs that perhaps I would be better simply fixing them in the old code. At least now I do have the VSCode stuff working and at a reasonable font size so I need to take stock on just what still needs doing on the new build before making a decision. I think that I probably have enough functionality in the new code to display all the websites as they are on the old platform and so it is time to start the switch. I should be able to run both in parallel anyway initially, perhaps with .uk on old code and .co.uk on the new. One thing I do need to get rid off is the crash that causes me to loose entries made to the blog pages as just happened. At least this time I HAD remembered to copy the new text before trying to save.

The priority stack that Claude has just summarised:

1. Get the installer working ✓
2. PHP 8.4 compatibility on Tags/Pigeonholes ✓
3. Modern style within Bootstrap 3 ✓
4. Bootstrap 5 migration → "later, if ever" ✓

That's pragmatic. You're modernizing the infrastructure (PHP, Firebird, Nginx) while keeping the interface layer stable. That's exactly how you should approach legacy system upgrades.

The MERG archive will work beautifully on this stack, and your users won't care whether it's BS3 or BS5—they'll just care that it's fast, searchable, and lets them find what they need.

The key here is I have been looking at the archive of material available from MERG behind their firewall, and seeing how I can make it more accessible. The starting point is for all of the pdf material to be searchable on the website, and a key element I have finally sorted out is displaying pdf files IN the page, with a working highlight direct from the search link. This makes navigation between pdf's a lot easier than when they simply open in the browser as a new page. Getting the find bar to open was something Claude distracted me from for two days, when the answer was actually very easy. It is just a pity that the CURRENT hack is not properly linked to on the pdf.js github threads.

Getting everything clean on PHP8.4 has been a pain, but all the live sites are now using the latest build of the code, and both pdf's and videos are actually displaying as the used to some time ago. Now to attack the installer and see if I can create a new site from scratch to play with the MERG material directly.

One of the reasons for wanting to get this sorted is to restore the search facility that bitweaver provides in the background, but while I could identify which pdf contained a search term, getting the generic viewer to do more than simply highlight it took a couple of abortive days with Claude. Not totally wasted time as I now have a much better appreciation of how to debug the javascript in firefox, but much of what was being documented actually worked at all! I have a large list of extra code which was intended to solve the simple problem of putting the search term in the find bar of the viewer so that the previous/next buttons worked. I had specifically asked about passing this in the pdf.js code but that path was avoided, however a github pdf.js issues described a fix that sounded perfect. The only problem was it's code is 9 years old and trying to find out how to apply it was not easy. Eventually I found that 'if ('search' in params)' is now 'if (params.has("search"))' on line 1250 of the viewer.mjs file. Another bit of fun since my nginx setup would not allow .mjs files to run ;) . But the next question is just what do I add to emulate the old example. This actually turned out to be easier than anticipated and the new code is.

I am considering if there should be a highlight param that just highlight but does not open the find bar, but at least I am now in control, and the bitweaver highlight param opens pdf files with the same highlighting as other directly generated content.

The NEXT step is to get uploading pdf's to automatically populate the liberty content object's data field with the pdf's text layer. It's working perfectly when I update existing pdf's, but I need a slightly different hack to allow that to happen on initial upload. That and getting bitweaver to actually index the pages with the search module. All the parts are there, and hopefully without AI assistance I'll get it sorted quicker!