How I got into Emacs (and Free Software)

This is a long tale, not confined to Emacs alone. This is, indeed, a story about why I am a confirmed Free Software advocate. This has been a rocky road. For this chapter, I'll try my best to restrict my discussion to Emacs; but that's impossible. In the late 80s and early 90s, I was living in Chuuk Lagoon, Micronesia. Graduate work at the University of Guam marine lab had been interrupted, at the outset of my third year; in Chuuk, my wife was pregnant, near term. In my letter of intent for my application to UOG Marine Lab, in 1984, I had written that a master's degree was not imperative, but my first order of business was the learn the marine fauna (and flora, I think) of Micronesia; I was preparing to study traditional marine biological knowledge in the islands. First, as I had related to Professor Thomas Harding at the UCSB Anthropology Department, I would prepare a lexicon of marine life. I had completed over two years of classwork, and had garnered sufficient knowledge of the fauna that, once I had arrived back in Chuuk, in late 1986, I hit the ground running, and started collecting notes, mainly on the names of marine animals, but also on the lore and knowledge that was an inevitable part of the package. Hopefully, later, I would interview experts: fishermen, navigators... whomever; that is when the real work would begin. Ad notes piled up, it seemed to me that a PC would be the most efficient way to compile, sort, and analyze the burgeoning information. I had a working knowledge of computers, and had used some of the earliest PCs at the Marine Lab. I had an inkling of an idea for typing in the names and organizing it with other data. I tried a few things, even purchasing a copy of SNOBOL4, and learning another language (I don't remember which), for string manipulation. It occured that this would be a useful way to approach and massage the data. I made a little progress. A software package that was somewhat useful was Borland Paradox. It is hard to find any information anymore, in 2026, about this software; it was kind of a combination database and spreadsheet program, where the data could be displayed in different models. Progress was, again, minimal. In 1990 and 1991, my life went through a time of turmoil. I separated from my wife, and my son and I moved to Fono Island for a year or so, where electricity was absent. My Mother came to the rescue with a Toshiba laptop. After it arrived by mail, I realized it came with no software, not even a word processor. I made do with whatever I could cobble together. I started using 4Dos, something of a DOS clone, but free software, or shareware, even better than DOS. I don't remember what software I used. I do remember falling in love with that OS, because a file listing could include comments, in addition to the usual MSDOS short filenames, making it easier to organize files. In the Summer of 1991, I had remarried, moved to Tol Island, and had a new job at Faichuk Junior High School. At some point I had been kindly offered a large 12V truck battery by the US Airforce Civic Action Team, which had to update batteries every year; and someone, somehow, gave me a small, 10 Watt solar panel. So on Tol, where there was not electricity, I would be able to hook up to a laptop charger (automotive type), and I rewired a portable HP printer, to run directly off 12VDC by tapping into battery leads. I also did the same for my Zeiss microscope. I had to take the battery to town, by boat, every three weeks for charging; meanshile the solar panel kept my computer, printer, microscopes, and a light going. I now had a laptop, and I tried various editors to develop a simple database---a list of animal names with descriptive text. Word incorporates control codes into the text, making a mess of things. I needed a programmer's editor. A linguist from the University of Hawaii (UH), Robert Hsu, sent a demonstration copy of MultiEdit, which did some of what I wanted. But it did not enable me to type diacritical marks. Chuukese vowel variants are typed using the diacritic "'" are á, é, and ó. MultiEdit would be able to do this, but to learn how, I would need to purchase the manual for over 300.00. I had nowhere near enough money to purchase this. About that time, I had been receiving a subscription to Infoworld, a computer trade newspaper that is available by free subscription. One day, a short article was buried a few pages in, about the Free Software Foundation. I knew nothing about this, or about Free Software, but the name said it all. I wrote a letter to the FSF explaining my project, and asking for help for an editor. A few weeks later, I received a package with about 13 3-1/2" diskettes, a package that would change my life. On one disk was "Demacs," an implementation of Emacs for the Windows ecosystem, in addition to a number of Unix text tools ported to Windows, by Cygwin. The unix tools like sort, ptx, grep, find, and others were helpful for my lexicon project. Demacs was a well-functioning port of Emacs to Windows. One of the first things I noticed about Demacs was the existence of a Tutorial, a genius introduction. Within minutes I was sufficiently able to navigate the interface and work with text. I developed a crude understanding of Emacs Lisp, and wrote several low level helper functions for my needs. This software was beyond any dream. I was working with textfiles, and since Emacs/Demacs is a programmer's editor, the files generated do not incorporate any formatting marks, which would interfere with manipulation of text, words, and characters. Shockingly, in this age of continual modernization, most of the tweaks I developed to my first.emacs file still function on today's Emacs 30+, in 2026. And, of course, since my project, and others, were generated in text files, for the most part, they are still readable today. This is not the case for files written on early windows Word versions. Happily, over the next couple of years, the Gnus Bull, the bulletin of the Free Software Foundation arrived every so often in my post office box. In one of these bulletins, fascinating reading for me at the time, was an article about Unix workalike operating systems, Linux and FreeBSD. Later on, on an extended medical visit to Guam, I was able, through generous assistance of the owners of Kuentos Communications, including Paul Carlson, I was able to download one of these and install it on my Toshiba Laptop. I randomly selected Linux, and I am certainly glad that I did. This was the Slackware distribution of Linux. This now meant I could install Emacs directly, and not a Windows port. The Unix text tools were available directly as well. The use of My needs are simple, compared to many---if not most---Emacs users, many of whom are programmers comfortable with generating elaborate and extensive code. As my project grew, Emacs was the essential ingredient. Later on I learned of the existence of LaTeX. Eventually, my lexicon was published from camera ready copy generated by LaTeX in the journal Micronesica. Many scholars and programmers helped along the way. Among them was Professor Stampe of the UH Linguistics Department, who helped, over an extended period of time, to develop a sorting program in Emacs Lisp to 'alphabetize' my list, including diacritics, in a unique order, including features to make it easier for an english speaker to search the list. Early on, Professor Hsu, mentioned above, introduced me to his proprietary SNOBOL4 and SPITBOL text manipulation tools that were widely used in developing the PALI series of dictionaries of Micronesian Languages at UH. Among the innovations I encountered were the BAND format, a form of free form database for freely structured lexicographical data widely used by linguists for word lists. I was able to implement this format in later stages to compile the database that was pivotal in the final compilation to LaTeX, and ultimately PDF formats. I was able to use both built-in and custom emacs lisp functions to manipulate this data. Also useful were the I wrote emacs lisp functions to scan the lengthy band formatted file and output latex source.

Setting file times in Emacs from Time-stamps in file, or "Created " at start of a file: Generated through Google AI Summary

AI is problematic, but the sheer convenience of it has grabbed my attention.  For tweaking my GNU/Linux system, AI summaries of Google searches come in a few seconds, and they sometimes save hours of research.  It is helpful that low grade links to sources are often provided.  I balk, however, at the damaging infrastructure involved.   

 A recent adventure has been developing a useful emacs lisp tool to change the metadata time stamps of a file from emacs' Time-stamps, when they exist.   

Today I realized I had copied files from another system without specifying to save file metadata.  Could I set the file times of files with timestamps using the "Time-stamp: < >" utility in Emacs?   I searched, and, I must freely admit, I asked for more info from the AI helper of Google Search.  After about 7 or more rounds of providing feedback about  a regular expression that did not work, these functions were provided.  Wow.   The last two of these is probably sufficient for my needs.  I will change the names.

 

(defun my/set-file-time-final ()
  "Sets file time by finding 'Time-stamp' and extracting only the date/time.  From Google AI."
  (interactive)
  (save-excursion
    (goto-char (point-min))
    ;; 1. Find the literal word "Time-stamp" anywhere (ignores #, brackets, etc.)
    (if (re-search-forward "Time-stamp" nil t)
        ;; 2. Look for the first date pattern (YYYY-MM-DD HH:MM:SS) that follows it
        (if (re-search-forward "\\([0-9]\\{4\\}-[0-9]\\{2\\}-[0-9]\\{2\\}[ T][0-9]\\{2\\}:[0-9]\\{2\\}:[0-9]\\{2\\}\\)" nil t)
            (let* ((ts-string (match-string 1))
                   (parsed (parse-time-string ts-string))
                   (lisp-time (when parsed (encode-time parsed))))
              (if (and (buffer-file-name) lisp-time)
                  (progn
                    (set-file-times (buffer-file-name) lisp-time)
                    (message "Success! Mtime set to %s (Username ignored)" ts-string))
                (error "Parsed '%s' but system couldn't use it" ts-string)))
          (error "Found 'Time-stamp' but no YYYY-MM-DD HH:MM:SS format found after it"))
      (error "The word 'Time-stamp' is missing from this file"))))

(defun my/dired-set-times-from-stamps ()
  "In Dired, run 'my/set-file-time-final' on all marked files. From Google AI."
  (interactive)
  (let ((files (dired-get-marked-files)))
    (dolist (file files)
      (with-current-buffer (find-file-noselect file)
        (condition-case err
            (progn
              (my/set-file-time-final)
              (save-buffer))
          (error (message "Skipping %s: %s" file (error-message-string err))))))
    (revert-buffer) ; Refresh the Dired buffer to show the new times
    (message "Done processing marked files.")))

(defun my/dired-set-times-from-stamps ()
  "In Dired, sync system times for marked files. Files without stamps are ignored."
  (interactive)
  (let ((files (dired-get-marked-files))
        (skipped 0)
        (updated 0))
    (dolist (file files)
      (with-current-buffer (find-file-noselect file)
        ;; Use condition-case to catch files that don't have the stamp
        (condition-case nil
            (progn
              (my/set-file-time-final)
              (save-buffer)
              (setq updated (1+ updated)))
          (error (setq skipped (1+ skipped))))))
    (revert-buffer)
    (message "Done! Updated: %d | Skipped (no stamp): %d" updated skipped)))

(defun my/set-file-time-final ()
  "From Google AI summary session after Google search.  Sets file time from header. Handles: Created YYYY-MM-DD Day HH:MM:"
  (interactive)
  (save-excursion
    (goto-char (point-min))
    ;; 1. Search for 'Time-stamp' or 'Created'
    (if (re-search-forward "\\(Time-stamp\\|Created\\):?" nil t)
        ;; 2. Match the date (YYYY-MM-DD) 
        ;;    Then skip any day-of-week (like Tue)
        ;;    Then grab the time (HH:MM or HH:MM:SS)
        (if (re-search-forward "\\([0-9]\\{4\\}-[0-9]\\{2\\}-[0-9]\\{2\\}\\)[^0-9]*\\([0-9]\\{2\\}:[0-9]\\{2\\}\\(?::[0-9]\\{2\\}\\)?\\)" nil t)
            (let* ((date-part (match-string 1))
                   (time-part (match-string 2))
                   (ts-string (concat date-part " " time-part))
                   (parsed (parse-time-string ts-string))
                   (lisp-time (when parsed (encode-time parsed))))
              (if (and (buffer-file-name) lisp-time)
                  (progn
                    (set-file-times (buffer-file-name) lisp-time)
                    (message "Success! Mtime set to %s" ts-string))
                (error "Parsed '%s' but couldn't use it" ts-string)))
          (error "Found keyword but date/time format was unexpected"))
      (error "No 'Time-stamp' or 'Created' header found"))))

(defun my/dired-set-times-from-stamps ()
  "From Google AI summary session.  This calls the function my/set-file-time-final.  In Dired, sync system modification times for all marked files
based on the 'Created' or 'Time-stamp' lines inside them."
  (interactive)
  (let ((files (dired-get-marked-files))
        (updated 0)
        (skipped 0))
    (dolist (file files)
      ;; Open the file in a buffer without switching to it
      (with-current-buffer (find-file-noselect file)
        (condition-case nil
            (progn
              ;; Call your working 'within-the-file' function
              (my/set-file-time-final)
              ;; Save the buffer to ensure metadata is flushed to disk
              (save-buffer)
              (setq updated (1+ updated)))
          (error 
           (setq skipped (1+ skipped))))))
    ;; Refresh the Dired buffer so you see the new dates immediately
    (revert-buffer)
    (message "Batch complete: %d updated, %d skipped (no stamp found)." updated skipped)))

(with-eval-after-load 'dired
  (define-key dired-mode-map (kbd "T") 'my/dired-set-times-from-stamps))

Shinestacker for focus stacking

 This is the first image I processed using Shinestacker.  It's a remarkable and technically advanced piece of open source software.    This software is beyond my skill level, but works incredibly well with defaults.  

 

Sufficeth to say it was very simple, after a somewhat convoluted---but not difficult---installation on CachyOS, to process this stack of 40 images.  I have a CPU with 27 or 28 processors; and this software even took advantage of my aging Nvidia GTX-1050 video card with a mere 3GB of video ram.  25X objective, mounted in UV-curing glass glue from a hardware store.  Too bad about the detritus.  Processing took a few seconds.  

 

Ossicles of Holothuria atra mounted in UV-curing glass glue.

 

Actually, Wayland is usable. I am leaning on Sway on my laptop.

 Just to bring the conversation up to date.

 

 

Wayland is not happening for me

Due to issues with HiDPI (High Dots-Per-Ince) on my Lenovo Thinkpad X1 Carbon, I wanted to try Sway and Hyprland, which were touted to handle HiDPI displays transparently.   The original issue was mainly that my go to i3 tiling window manager does not pay well with HiDPI resolution.  Text was unreadable for some apps.   

 It is true.  Out of the box, Sway---especially installed as a complete distribution of Manjaro Sway Edition---worked nicely.  Many key bindings were compatible with i3.  In a day or two, I was up and running.  

Along the way, I tried Hyprland, an overly decorated tiling option.  Hyprland works really well.  It was able to run applications like RawTherapee that did not work well for me on the Sway install.  Even the flatpak of Entangle Photo Manager---an essential package for me---worked well.  Not so for Sway.  Today I found an application that does not work with Hyprland, either.  

 I have maintained i3 on my desktop linux box.   It seemed ok to use either of the two Wayland options on my laptop; but jumping from system to system is too painful, when some apps don't work on one, but work fine on the other.  

 Wayland has some downsides, imho.  Configuration is a little too nerdy.  

 When wayland first arrived on the scene, I wondered what could be so bad about X11 that these changes were needed.  I'm getting tired of jumping through update hoops; that's why I quit Gentoo a long time ago.   I now learn that Wayland, slick as it may be, is actually not a drop-in replacement for X11, which I am now used to.

 Apparently Xwayland, often mentioned as a way to run X11 applications on Wayland machines, is not as simple to use as promised by Sway users.  There are some other issues, that apparently X11 compatibility do not solve.  I have not figured out how to configure Xwayland on Sway.   

 Goodbye, Wayland, at least for now.   Maybe I'll try you again, someday.  But why?

 

 

Trying out Sway, an i3wm workalike for Wayland

 A good move I recently made was to sell my MacBook Pro, M2; and purchase a Lenovo Thinkpad X1 Carbon, several years old.  One of the best features of this new laptop is a HiDPI (High Dots Per Inch) screen with a resolution of something like  3200x1800.  It's a remarkable display for other reasons, which I will not go into here.  The blacks are incredible.  

 With my usual Manjaro i3wm, this means that the type is almost unreadable.  With the guidance of several source on the Internet, I was able to cajol the system into somewhat useful settings that most software works fine with.  Along they way, several references passed me by about Wayland, a newer graphical setup that X11.  

 

I am reluctant to make huge changes to a long time stalwart system, but finally I have begun to experiment with Sway, a workalike system to i3, and Hyprland, a fancier tiling window manager, working on top of Wayland.  Both, I am happy to report, work nicely with this  HiDPI screen.  

 

So I installed Manjaro's Sway edition.  So far, so good.  Now I'm at the point of trying to configure pop-up/drop-down floating windows for Kitty, Emacs, R, orage (probably impossible, but this is the best little calendar app I have found.  Once I have set the month, and popped it back up, the same month is show.  Very useful for calendar work), and a ham radio clock, that is actually of little use.  

HiDPI is not a problem.  

 

When I install a new distro, one of the first things I do  is set the user id to "1004", as well as the groupid.  Consequently, the home directory of my previous setup and user are completely accessible, because this was the case for the previous login as well.  I never duplicate the login, out of worry that the dot file tweaks will be incompatible with anything different.  

 

So far, so good.

 

Tools of Excellence: Cb2bib, Yazi, Sioyek, Entangle, XnViewMP, Emacs,...: WOW!

In the summer of either 1982 or 1983, I enrolled in a summer school class at UCSB offered by the College of Engineering on Computer Architecture.  At about this time two of my regular classes required the use of Wordstar, on a mainframe terminal.  I had been required to write a program in Fortran for a math course; now, in this Computer Architecture class, I wrote some minimal code in PDP-1170 Assembly Language.   Intel's early microprocessors---the engines that  would spark a revolution in Personal Computing---had come into being.   I had enrolled on a Pass/Fail basis, fearing that it would be a technical challenge.  I had enrolled later in an amazing  Invertebrate  Paleontology course, also Pass/Fail.  Both times, I would have earned an honest "A."         Yet another Lesson learned.  

 The theory behind digital computers, as it turns out, is straightforward and fascinating, logical and understandable.  I later came to realize that these systems, invented  by humans, reflected human logic; they are non-complex, a fact I understood more fully after being introduced to Molecular Biology and Genetics, and the extreme complexity of the Central Dogma  By the end of the short summer session, I had a tentative grasp of the nature of digital computers.  The graduate student instructor took me aside and advised that the knowledge I had gained in this brief introduction would be valuable in the future, and I would be able to advise others about what computers can do.  

 This advice could not have been more applicable when I moved to Chuuk Lagoon, and started a project collecting animal names in local dialects of the islands.  At the University of Guam Marine Lab, I was exposed to early Personal Computers, the IBM XT and IBM AT.  My basic knowledge of the workings of computing systems provided useful perspective for solving problems.  I eschewed the commercial/proprietary solutions, the high priced computer programs that would make the entrepreneurs wealthy.  

 I learned what I could and scrounged software of many kinds.  Except the big brands.  There are always other solutions that work better; those behemoths aren't worth the cost to society.  I believed this then, and I believe it even more today.  

 My Toolbox

Previously, I have written a little about Yazi and Sioyek.   They, like Cb2Bib and Entangle, are functionally remarkable tools.   Among Free Software users, they are not the most popular.  I also rely upon the highly functional, low overhead foundation, the i3 tiling window manager.  Emacs is a mainstay; like these other tools, the simplicity and functional design of Emacs makes it run circles around most proprietary software.  Windows is dumbed down; the stress of using such a dysfunctional tool might be likened to trying to build a cabinet using a jackhammer; a poorly made jackhammer at that.   I use LibreOffice from time to time, and some other more mainstream tools.  They are fine; but the workhorses in my toolbox are more special. The packages I am writing about today are truly special, and enabling. 

 

I as initially attracted to Free Software for financial reasons.  The story is too long to recount here.  But as I have come to understand the GNU Project, and the grass-roots nature of the development of these tools, it became clear to me that the Free in Free Software actually refers to freedom, not free beer.   

 

Sioyek

The PDF reader Sioyek caught my attention relatively recently, with its plain exterior, focus on research, and its  hidden gems.  For a couple of days, I returned to sioyek, after setting it aside a while; a closer reading of the documentation has given me a new appreciation of it's wonders.  And today, after some digging, I was able to configure yazi to open files in Sioyek, or Evince, at will.  Gradually, the levers and buttons of sioyek are revealed.

 

The documentation is relatively straightforward.   But I struggled with configuration.  Later for that.  PrintHere are some of the more interesting features:

• A keystroke sends the text at the cursor to Google Scholar (GS), or another tool I am unfamilar with.  A browser opens, Google Chrome.  Keystrokes are available, either standard or by a plugin, to download the pdf from GS.  
• Sioyek remembers bookmarks of two or three kinds.  A keystroke causes display of a list of bookmarks in either the current file, or all pdfs sioyek knows about.  
• One of these "bookmark" functionalites is called "highlighting."  Selected text may be highlighted.  The keystroke for highlighting must be followed by a single key, which represents the type of material.  So someone like me, whose mind travels in so many directions, can keep track of various threads of interest.  If several files are highlighted in this way, retrieval of highlights can optionally display all highlights of a given type in all files (globally).  
• The Global retrieval feature holds for bookmarks as well.  So if I am reading four PDFs that relate to each other, I can bounce around.  And selections can be copied to the system clipboard.  You know where this is going.
• Portals are something I don't yet understand.  I position may be set as a target for a portal, so as I'm reading I can refer back to a certain specific piece of text, diagram, table, whatever.  Backspace brings me home.
•  I was able to print.
• Sioyek is mainly keyboard driven; but scrolling with a mouse is possible.
• A "visual mark" (or whatever it's called) underlines current text if that is wanted; and this mark can be manipulated in various ways.  More to learn hear.
• Zooming is great.
• In PDFs text can be selected and various functions are possible.  A click on a figure number will take me directly to that figure.  This built-in feature seems almost intelligent, and it reminds me of snarfing references in Cb2Bib, a remarkable feature.  
• Synctex mode allows the latex source to be opened and followed when reading a pdf.  Wow.

Some thought went into this tool.  

 

Yazi 

Yazi is a console based file manager with superpowers.  I have gotten into the configuration details far enough to make it possible to choose which pdf viewer is desired.  Sometimes, I think sioyek may be too far out, for some things; so I set up to be able to open in either  evince or sioyek.  Shift-O opens a minimal menu; included here is a button to open it in a folder.  I think some more tweaking is needed to identify low-overhead tools like qutebrowser for Google Scholar searches, and a minimal Graphical File Manager with lower overhead.  My laptop lags seriously, but it's a very old laptop.

 

Yazi works swell with kitty.  Previews of jpegs and pngs, and possibly tiffs (with some tweaking) is extremely fast, and a plugin "zoom" feature enables the image to zoom to the full height of the window.  Image and PDF previews are useful.  

 

Bunny is an amazing bookmarking tool.  

 

A tweak in .bashrc (or .zshrc) makes it possible to open yazi by typing 'y'.  As a bonus feature, when one changes directory, then types "q" to quit, yazi leaves the console in the last folder visited.  

 

This tool is undergoing rapid development, and plugins have been known to fail after updating.  

 

I use this for editing filenames, with tags.  A filter function ("f") then makes possible isolating files with a specific tag, marking them all, creating a directory, and dumping all these into a directory.   

 

Cb2Bib

 

This is an extremely useful tool, to me.  I think I'm the only one who uses it in the way I do, to write annotations to records in my bibtex database, and print out annotated bibliographies on demand.  I write, for example, a call number of a book, and notes on pages, compile it into LaTeX, and print.  Very convenient.  The developer was kind enough to explain how to modify the code to incorporate an "annotate" field in the bibtex entry, and also modify the tools that generate a pdf, so an annotation is formated together with the citation.  

 

In this space, I hope to present instructions for the benefit of others.  

Cb2Bib also automatically picks up citations that are highlighted in other programs.  Also, one can copy a bibtex citation from Google Scholar, that is instantly picked up and incorporated into the database associated with the program.  

This program does much more than I am able to understand, but what I can understand, and do---with the help of Pere Constans---is amazing.  

 

This is another reason I don't stick to the commercial / proprietary software packages.

 

Entangle

 

More to follow. 

 

 

Tangled in Entangle: resignation to flatpak

Entangle is a wonderful program, an elegant tool---elegant in the sense of efficiency and design for function.   Daniel Berrangé  has written other uncommonly interesting  software as well.  This tool, which I run on GNU/Linux, flawlessly connects to my Canon DSLR cameras; I use it primarily for photomicrography.  What it lacks is documentation.  The GUI is not super shiny, but adequate.  

I have taken Entangle for granted for many years; it has installed easily on GNU/Linux Arch-based systems---via the AUR repository---and also on Ubuntu, as I recently learned.  However, for several months it has been a struggle.  On Manjaro GNU/Linux, my goto distribution for over a decade, building has been impossible, for me.  Because Entangle is so important to me, I moved to EndeavourOS, a distribution that otherwise is not my favorite; Entangle did build fine on EndeavourOS, however.  But recently, other issues plagued my EndeavourOS experience, initiating a devastating avalanche: my desperate flailings led to the loss of a 600GB partition, that had not been backed up except in some bits and pieces, with decades worth of work, now completely unrecoverable.  

Monofilament fishing line sucks.  Google's AI suggest some terms for  snarls of monofilament on a reel: backlash, bird's nest, webbing, or wind knot.  I have forgotten the more colorful term I knew long ago---explosion?  I had to cut the mono off of the reel.  (I no longer fish, out of concern for the health of the ocean, and overfishing).  This week's avalanche is another lesson, because the entire journey was unnecessary, had I but taken heed of the notice on the web page of Entangle: flatpak is the method of choice for installing Entangle:

 

At least I can now report that I am back to Manjaro.  Each of the other distros had other shortcomings; so does Manjaro, but in the main, it is most comfortable.

 

EndeavourOS

As a firmly committed user of the tiling i3wm, this distribution is clearly out of bounds.  I3 has been over-riced and the keybindings are out of sync with any others I have seen.  As one reviewer remarked, Endeavour''s setup seems more like a single user's custom setup than to reflect any functional vision.  I love it uses Archlinux's upstream repos (this is the most serious issue I have with Manjaro).  I can tweak it to my satisfaction, but the theming is garish and overdone.

 

Later this week, I attempted to install EndeavourOS again.  But the web site is down, and I was unable to download the install image any longer, or find the wiki or forum.  I think this is the death  knell for EndeavourOS.  It "couldn't have happened to a nicer distro," but it's sad that the reason turned out to be inability to pay hosting fees.  

 

Don't get me started about Capitalism. 

 

CachyOS

On Distrowatch CachyOS---a newer Arch-based distro---has snuck into first place, quick swiftly.  Reviews are quite positive.  The vision is amazing; a breath of fresh air.   I took the bait, and grabbed an iso.  The installation went well, or so I thought.  I even let out a joyous laugh, while installing it.  (Someday, this will be my fave.)  This is the ultimate functional distribution, even including a stress testing facility on the installation medium.  But when I booted the installed setup... crickets.  Black, blank screens.  After too much sleuthing---this past week has been all about too much sleuthing, no working---I found a plethora of complaints about this issue, on Nvidia GPUs.  The issue for me, as it took days to figure out, is that the GTX 1050 of mine is no longer supported by Nvidia; but it works on all other distros.  CachyOS has favored a newer open source driver (which I'm all about, but this new open source driver is not Neuveau.  It's apparently an officially sanctioned open source driver, but older hardware is still closed source, so are not supported.  

 

Curiously, once, after I yanked the HDMI cable to one of my monitors, test output appeared on the other monitor, connected by display port.  I noted a single mention, during my sleuthing, of the issue with HDMI.  

 

For now, this is the GPU I have  So no more CachyOS.  Too bad.  It was a real thrill to install something so fresh and cool.  I tried Manjaro again, but no luck with Entangle---and the module/drivers for my USB Wifi Dongle was not included with the 6.12 kernels, so I tried something new.  Or old.  The hard way. 

 

 

The Semi-Hard Way: Installing Archlinux

All of these "Arch-based" distros are fed by Archlinux itself.  I have installed Arch a bunch of times; and used various short-cut scripts and (of course) the arch-based easy way distros, like Manjaro, Antergos, and others.  I have done it enough times, the hard way.  On the forums, developers of Arch make much ado about the need to install the hard way as a way to learn GNU/Linux; but enough is enough.  The General Purpose Computer is a tool; I want to understand this tool, sure, but that is not my raison d' etre.   

 

Anyway, at this point in the trip down the abyss, I thought, well, why not go ahead and reach for the progenitor of these others, settle down, and bite the bullet.  I used the "somewhat hard" way.

 

Arch's installation iso images now include a script called Archinstall (or so), that is intended to make it easier to install.  I experienced a disaster of untellable proportions while using this script.  I had forgotten this, but some years ago, apparently, I had attempted to use the script, and had been stymied by the disk partitioning module.  Once something has been typed in to the widgets, indicating mount points, it is difficult, or sometimes impossible, to back out of it.  In this instance, I encountered other problems with the partitioning component of this script: I had mostly figured out the correct paritioning scheme, and eventually had them all in place, but once I had typed something into the wrong partition, it was impossible to back out, so I typed "/extra".  Who knows whether that affected our issue. 

 

The notice escaped me during the process that the partitions that were selected for the install would be wiped.   Somehow, it occurs to me that this notice should not be a subtle one.  I cannot think of another distro's install that does not take extra care to explain the consequences of similar actions.  

 

Not only that, there is a button, not clearly defined or marked, for each partition defined, to change the formatting status (ie, format as opposed to keep).  It is not only poorly marked: when I tried to install again using this script, it was clear that this button only switches to "format" and does not seem to convert back to a "keep" state.  The wording was difficult to understand, for me.  This, I think, is a bug.

 

Immediately after I clicked "install"  I saw the words "partition deleted"; I killed the install at this point.  Once I had taken a look at the system, using gparted, it was clear that this partition had been deleted and wiped.  Nothing I could do could possibly recover my data.  This only reinforces my sense that this script reflects a disdain for the user, so, arguable, it is a bug.

 

At this point I decided  to install Ubuntu.  I have had several problems with Ubuntu in the past.  Installs that led to blank screens on laptops.  But it's been some time.

 

Ubuntu

Ubuntu installed nicely.  It was a breath of fresh air.  But it's tedious to maintain and set up.  Many of the packages I install routinely on Manjaro are difficult on Ubuntu.  Entangle did install nicely.  

 

Final Chapters

At this point, perhaps four days into this waste of time, it occurred to me that maybe a binary package was available on the Manjaro web site.  What I found was the notice, above, that the canonical way of installation would be flatpak.  

 

Back to Manjaro.  What's to complain about?  It worked.

 

In Praise of Manjaro

 The issue with Entangle was a mirage; it does work, but requires installation as a flatpak.  I don't like this; hopefully in the future a new version will be released (it's been a long while) that will compile on today's systems.  But I can think of few or no issues over at least 15 years since I started using Manjaro.  

 

One think points to a high level of knowledge among the developers of Manjaro.  I have made it a habit to keep a separate /home directory and even /usr/local---for self-compiled software---so that if an issue does creep in, a quick installation of Manjaro to a different root partition will solve the problem.  Grub uses os-prober to locate other installed GNU/Linux sets on the same machine, and configure the grub startup screen to offer choices.  Only in Manjaro, once I have chosen a different partition to start up, does Grub remember.  This is not a unique deal; Grub makes this possible with a simple configuration tweak.   But this points to a level of attention to detail that is lacking elsewhere.

 

I might also mention i3wm.  I have a 2014 Thinkpad that still is a bit spunky, when not held back by an excessive amount of overhead from, say, KDE, Gnome, or even XFCE4.  It's a big deal.  My desktop is getting along in years as well; I think without i3 (or some other super lightweight window manager) I would have felt bogged down.  Not only that, i3wm is functional.  All of the ricing, all of the beautiful desktop tweaks and wallpapers, cannot compete with the beauty of a smoothly functioning machine. Also, i3wm is easy to configure, and some experts have provided great tweaks.  I have configured popups for R, emacs, kitty (terminal), and other tools.  Wonderful.

 

Manjaro, as an Arch-derivative, comes configured to take advantage of the AUR: Arch User Repository.  Using the tool "yay" (replacement for pacman), there are few packages that are not installable out of the box.   AUR packages compile programs from source, downloaded from git.  This is consistent with my inclination toward using programs in the manner in which they were written; Debian and Ubuntu, for example, provide patches for almost every package that alters the software in various---usually innocuous---ways.  

 

And Manjaro comes, out of the box, with development tools installed, so that for many programs, compiling from source is accomodated.  

 

I am pleased that I can now run Entangle, even if I must (drat!) resort to a flatpak.

 

One thing I do not like about Manjaro: the "repostories" lag behind those of Archlinux, the canonical upstream source repository, in the supposed name of security.  I like to run either the Testing or Unstable branch of Manjaro (I think one of them may not longer be supported), which is one step more up-to-date with Arch.  I blamed this for Entangle not compiling, but it appears I was mistaken.  

 

 

  

sioyek and yazi

I have set Sioyek as the default pdf viewer for Yazi; actually, probably for the entire system.  

 It was *relatively* simple to do this, after installing xdg-utils-mimeo.  I will not go back through that here, but it's not hard.  

 Sioyek has no native capacity to print.  This is a potential show stopper.  I have attempted to set this up from instructions from some random internet post.  The location of the user config files is in doubt, however.  By "reports," config files should be found in ~/.config/sioyek; however, other reports indicate .local/share/sioyek.  I did install a rather decrepid tool, a gtk-print tool (name?).  My config does not work yet.

Another potential downside relates to the use case for Sioyek, as a pdf annotation tool for research.  When a file is opened for the first time, sioyek indexes the file; this makes searches incredibly quick and useful.  Will this result in cluttering the system after a large number of pdfs are opened?  Probably another lightweight pdf viewer should be established as the default, to streamline the process of reviewing random files on the fly.  

Sioyek is uncluttered, as far as GUI.  There hardly is any GUI.  A fairly extensive set of utilities is tied to keystrokes.  Some of them are not exactly intuitive, at least to me.  

Some facilities:

• Bookmarking
• Search in google scholar (may not work right)
• gg -> go to page 1
• G -> go to last page
• rt click -> mark line (not sure this is working with a track pad)
• highlighting in various colors
• search highlights
• "Smart Jumps" to figures or references, with or without links: this didn't work out of the box, for me.
• mark a location, give character as a name for the mark.

 

Yazi is turning out to be extremely useful.  I use it multiple times a day.  Some useful tweaks:

• set a much larger preview size, then
• install a preview to Toggle full view of the preview
• somehow a special tweak was needed to view tiffs (this wasn't the case when I first tried yazi)
• bookmarks.  yabm (Yet Another Book Mark) is great.  Another one was incompatible, AFAICT.
• Zoom plugin
• In a tips section of the web site, is recommended to install some code in the rc file of the shell, which allows to start with "y" and that leaves the terminal in the last used directory, when quit with "q".  Extremely useful.

A problem: not really a big one: yazi is undergoing rapid and intense development, so incomptibilities arise frequently.   Especially with plugins.

 

 

Some recent useful software additions

 A brief list for now:

• dysk : concise "df" and graphic
• sioyek: interesting  document reader.  Keyboard driven, mostly, maybe. Designed for academic papers and textbooks.  
•  bpytop: A feature-ful process/memory/cpu monitor, text based; similar to top or htop.
  
•  

Notes on Yazi

 Yazi is good.  Real good.  I used it over ssh to cull a directory on my second system; it worked as I had hoped, except, perhaps, for color theming.

 Knock on Wood.

 Viewing/reviewing

One of the needs that yazi has fulfilled for me is rapid viewing.  I can only imagine that should I ever be able to upgrade my current machines (RAM and CPU, as well as Storage), viewing would be even more friendly, especially for images.  The ability to flip through a directory of images, mark, and move, copy, or delete marked files, has made a huge difference to my workflow.  The only other tool that is useful in this regard is xnviewmp.  I use a fairly simple set of commands; probably if I mastered tabs and other complexities and plugins, yazi would be even more helpful.

 

By the way, the terminal "kitty" has wonderful facilities built in to work well for viewing images.   I prefer the black background and contrasty color scheme to any other.

 Ranger was good enough, until it wasn't.

Copy, Paste and move and Notes on plugins.

 [Cautionary note: I have lost files through carelessness here, and yazi has some gotchas to be navigated around.]

This is pretty much self-explanatory.  I will try to address this topic in detail at some future time.  For now, I'll mention a plugin that has helped speed these processes: 

• yamb is a bookmarking plugin that works well.  In the absence of a much needed multi-step backup command, this does a good job when sorting through files and moving or copying them to another directory.  I have gotten around the mentioned limitation by defining  bookmarks for the current from directory, and the currently interesting goto directory,  I define a set of book marks to frequently used directories, as well.  This is an easy tool to master.  Like other plugins, issues have been experienced after updating yazi (it is under fairly intense development).  

• toggle-pane: This helps when viewing both pdfs and images.  It is possible to configure the mazimum size of displayed images (I'm not certain whether this works for pdfs) by putting this code into ~?.config/yazi/yazi.toml:

                       [preview]

                      # Change them to your desired values
                      max_width  = 1800
                      max_height = 1800
  

       Plugins are found on a resources link on yazi's web pages.  Other sites exist with plubins.

 

Filtering

Filtering is a wonderful way to sort.  It has been even more useful to me, because of a filenaming scheme I have adopted.  I stole this idea from Protesilaos's Denote system.  I have not found denote very useful, but the concept of tags at the end of the filename, before the extension has become a useful tool allowing me to sort files more rapidly.  Here is an example.

    Jones--2060--JargonInBiology__taxonomy_nomenclature_zoology.pdf

I have started avoiding underscores in filenames ("_") in favor of using them in tags.  Interestingly, I saw a web page or thread somewhere about tagging styles, and this is one of several.  I used to use a 10 (or so) character at the beginning of lines in a bibliography, where each position had a meaning.  So anything related to crabs would have "c" in the third position (12ctx8urt-Anon-Crabs of the world) would be sorteable by sorting by the third position.  The Unix (GNU/Linux) sort command worked excellently.  

 With filtering  in yazi, one can narrow down a large folder/directory with pdfs about taxonomy: "_taxonom".  Prot has demonstrated the use of regular expressions to sort even more specifically in, I think emacs dired.  To avoid filtering for "_" in filenames, one would theoretically use a regular expression requiring a "__" (double underscore) somewhere before the "_:" single underscore.  This is probably easy.  Filtering is easier, I think.  It may be that emacs would be better, but so far, aside from a couple of learning experiences, Yazi is working well.

Moving around 

In either the getting started or tips section of the yazi main web pages, is a suggestion that makes yazi more useful.  

Inserting this code into .bashrc (and I think .zshrc as well) makes things easy.

 function y() {
    local tmp="$(mktemp -t "yazi-cwd.XXXXXX")" cwd
    yazi "$@" --cwd-file="$tmp"
    if cwd="$(command cat -- "$tmp")" && [ -n "$cwd" ] && [ "$cwd" != "$PWD" ]; then
        builtin cd -- "$cwd"
    fi
    rm -f -- "$tmp"
}

Then, when in a terminal, type "y" to start yazi.  After moving to another directly, quitting with "q" leaves the terminal in this new directory.

 

A Caution or Two 

 Plugins are often not mutually compatible.  I have tried several others, but some collided with these few i do use.  

Some plugins iexplicably use keybindings that are assigned to other functions by default, and it is not always apparently what key bindings may be used.  <F1> may display a help screen, which may be filtered.  Filtering rocks.  Bookmarks rock.

 

2025 May 19