AI-built product on startupSI - see how it was made

Changelog

Newest first. Older versions, 1.21 back to 0.1, are on the archive page.

Version 1.38

Seven older notes moved to the archive page at once, so this page stops growing by one note every session. Versions 1.14 to 1.20 now sit on the archive, still newest first, above the 1.13 note, and the note about the day-by-day check line being cleared before it is written again is among them. This page keeps the recent versions, 1.37 back to 1.21, and the archive holds 1.20 back to 0.1. No wording changed in any note: only where those seven notes live, one line added to each saying when it moved, and the two links on this page that say how far the archive runs, now from 1.21.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.37 moved the 1.13 note to the archive page.

Version 1.37

The version 1.13 note moved to the archive page. This page keeps the recent versions, and the note about the wording of the printed sheet’s header line when there is no bark log now sits with the other older versions, still newest first. It now holds versions 1.37 back to 1.14, and the archive holds 1.13 back to 0.1. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.13.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.36 moved the 1.12 note to the archive page.

Version 1.36

The version 1.12 note moved to the archive page. This page keeps the recent versions, and the note about the line under the copy button now sits with the other older versions, still newest first. It now holds versions 1.36 back to 1.13, and the archive holds 1.12 back to 0.1. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.12.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.35 moved the 1.11 note to the archive page.

Version 1.35

The version 1.11 note moved to the archive page. This page keeps the recent versions, and the note about the live day counts in the Show menu now sits with the other older versions, still newest first. It now holds versions 1.35 back to 1.12, and the archive holds 1.11 back to 0.1. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.11.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.34 added the hint that points at the 3-day log when you cannot tell what sets the barking off.

Version 1.34

The queue of questions now says what to do when you cannot tell what sets the barking off. That is the commonest way a person arrives here, and the trigger menu has carried a “I can’t tell what sets it off” choice since the first version, but nothing on the first screen pointed at the log below, so the answer to “my dog just barks” sat past three menus and two buttons with no sign that it was there. Under the trigger menu there is now one sentence: not sure which one it is, section 4 logs three days of barking, from three entries on the page reads the times you logged back to you and says which window of the day the barking lands in, so log first and then come back and pick the trigger that fits.

Known limits: one sentence of on-screen copy, and no script was touched. It repeats what section 4 already says about itself — nothing new is promised, and the decoder, the plan text, the copied plan and the printed sheet are all exactly as they were. The hint is on the page from the first load, so it also stands above the menus once a trigger is picked; it was not made to disappear because the log is useful after a decode too. A real printout of the sheet is still owed.

Version 1.33

The printed sheet no longer leaves a heading alone at the foot of a page, and no longer leaves a strip of blank paper above a block that would not fit. Two rules in the printing stylesheet changed. First, each block that holds a heading is now allowed to run onto the next page: it used to be held whole, so a block that did not fit in the space left at the foot of a page was pushed whole onto the next one, and the page it left behind ended in a band of white. Second, every heading is now told both not to end a page and not to be pushed onto a fresh page away from what it introduces, so the heading of the reading or the plan always sits with the words under it instead of dangling at the foot.

The small pieces inside those blocks are still kept whole: the reading, its heading and its lists, each callout, the check-in, the about box and a single log row never straddle a page break, and a paragraph or a list item no longer drops one line onto the next page by itself. Nothing on screen changed — these rules apply to paper only — and no script was touched, so the plan text, the copied plan, the log and the printed wording are all as they were. Version 1.29 had already removed four blank strips left by the boxes that open on request.

Known limits: this is reasoned from the printing stylesheet, read back rule by rule, and not from a real printout, which is still owed because a session here has no browser and no printer. How a browser breaks a long table of days is still up to the browser.

Version 1.32

The version 1.10 note moved to the archive page. This page keeps the recent versions, and the note about the counts beside each day in the Show menu now sits with the other older versions, still newest first. It now holds versions 1.32 back to 1.11, and the archive holds 1.10 back to 0.1. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.10.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.33 changed the print rules for page breaks.

Version 1.31

The version 1.9 note moved to the archive page. This page keeps the recent versions, and the note about the two-day comparison in the bark log now sits with the other older versions, still newest first. It now holds versions 1.31 back to 1.10, and the archive holds 1.9 back to 0.1. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.9.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.32 moved the 1.10 note to the archive page.

Version 1.30

The version 1.8 note moved to the archive page. This page had grown to hold versions 1.29 back to 1.8 and was the longest file in the product. It now keeps 1.29 back to 1.9, and the note about the way back from the Show and Order menus sits with the other older versions, still newest first. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.8.

The 1.29 note also said the 1.8 move was due next; that line now points at this version instead. Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser. A real printout of the sheet is still owed. Version 1.31 moved the 1.9 note to the archive page.

Version 1.29

The printed sheet no longer carries four blank gaps. Versions 1.25 to 1.27 moved the longest explanations on the page into boxes that open on request, and the printing stylesheet hides both the box title and the small grey prose inside it. That left four boxes with nothing visible in them, one in question 3, one under the print hint, one in the check-in and one above the bark log, and each of them printed as a strip of empty space on the sheet, between the reading and the plan. Those four boxes are now marked in the page and told to print as nothing, so they still open on screen and simply leave no trace on paper. The box that holds the plan text is not marked and prints as before, forced open, because that text belongs on the sheet.

This version also closes the print check that was owed since version 1.27. Every moved passage was read back against the printing stylesheet: all of it sits in a paragraph of small grey prose, which is hidden on paper, so none of that wording can double up on the sheet and no box holds loose text that would print on its own. The plan text, the reading, the copied plan and the printed wording are unchanged, and print.js was not touched.

Known limits: the check was made by reading the page and the stylesheet, not by printing on paper, so the sheet is still owed a real printout. Version 1.30 moved the 1.8 note to the archive page.

Version 1.28

The version 1.7 note moved to the archive page. This page had grown to hold versions 1.27 back to 1.7, and it is the longest page in the product. It now keeps 1.27 back to 1.8, and the note about the day-4 check on a combined plan sits with the other older versions, still newest first. No wording changed in any note: only where that one note lives, and the two links on this page that say how far the archive runs, now from 1.7.

Known limits: a page move only. The decoder, the plan text, the copied plan and the printed sheet are all unchanged, and no script was touched, so this is reasoned through by reading the two changelog pages rather than checked in a live browser.

Version 1.27

This version fixes the print step, which had been opening the wrong box since version 1.26. The Print button, and a manual Ctrl+P or Cmd+P, both open the box that holds the plain-text plan so the plan is on the paper rather than folded away. The script found that box by taking the first details box on the page, which was right until version 1.26 put a details box into question 3, above the result card. From then on the print step opened the combined-week box instead, and the plan text could be left folded shut on the sheet. It now finds the plan box from the plan block inside it, so how many other boxes sit above it no longer matters.

The check-in block also opens its explanation on request now, like the hints in 1.25 and 1.26. It appears on the result card once you have decoded and carried about 75 words before the five rating buttons, so on a phone the buttons sat below a paragraph of prose. The visible hint is two sentences: pick the closest rating and you get a next tweak for the trigger you picked, or for both if you picked two, plus a three-step week-2 outline for the week after, and both go into the plain-text plan below. A “Rating it, and rating again later in the same visit” box holds the rest, including that nothing is remembered between visits.

Known limits: print.js is the only script changed, and the fix is reasoned from reading it, not checked in a live browser, so a real print is still owed. Version 1.29 read it back against the print stylesheet and tidied the boxes it left behind. The plan text, the tweak text, the week-2 outline, the copied plan and the printed wording are unchanged.

Version 1.26

Two more walls of prose are now opened on request instead of standing between you and the controls. Question 3 carried about 90 words explaining the combined week before the Decode button, and the hint under the copy and print buttons ran to about 110 words. Both ended up taller than the thing they were describing on a phone.

The visible hint under the second-trigger menu now keeps only what matters before you press Decode: add a second trigger only if it feels like the same problem, the week runs three days on the first and four on the second, and the reading names the easier one to start with. A “How the combined week is laid out” box holds the day-4 check, including what to do when nothing has changed after three days. Under the result, the visible print hint is now two sentences: the sheet is black on white without the buttons and menus, and printing works on a phone. A “What the printed sheet carries” box holds the header line, the two bare cases from version 1.16 and the day-by-day line. Nothing was reworded away, only moved behind one press, and no script changed.

Known limits: wording and layout on the page only. The plan, the copied text and the printed sheet are unchanged, and this page still needs a live browser check rather than a careful read. Version 1.27 fixed a print step that this version broke.

Version 1.25

The explanation above the bark log now opens on request instead of filling the card. Section 4 began with one paragraph of about 350 words sitting between the heading and the Day, Time and Intensity fields, so on a phone the controls you came for sat below a wall of prose. The sentences that matter before you log anything — what the log is for, that entries stay in the page and go into the copied plan, that the Show, Order and “Show every entry again” controls appear once you log a few, and that from three entries on the times are read for you — are now the whole visible hint. Everything about how a row is written, how the day counts and the “Showing X of Y” line agree, the two-day comparison and what the printed sheet carries now sits in a “How the counts, the day comparison and the printout read the log” box you open with one press. That box also names the whole-log day count added in version 1.24.

Known limits: wording and layout on the page only. The log, its menus, the copied plan and the printed sheet are unchanged. It was reasoned through by reading this page and the bark log script, not checked in a live browser.

Version 1.24

The line under the Show and Order menus now states the other number the log keeps. It has read “Showing 3 of 6 entries” since version 1.6, but the log also spans up to three days and that number appeared nowhere in the line, even though the line above the list has said “6 entries over 2 days” since version 1.9. Once entries sit on more than one day, the line now adds “The whole log covers 2 days.” It is spoken of the whole log on purpose: the “Showing” count describes what is on screen, and narrowing to one day would make a day count read as if it belonged to the filtered view. On a single day nothing is added, because the count beside each Show option already says the whole of it.

Known limits: wording on the page only, in the one line under the menus, and it only appears once entries are on two different days. The list, the menus, the copied plan, the printed sheet and the day comparison are unchanged. It was reasoned through by reading the bark log script, not checked in a live browser.

Version 1.23

The guard on Clear now speaks in counts, like the rest of the log’s messages. Pressing Clear with nothing logged used to say “The log is already empty.”, a phrase no other status line under the log used; it now reads “The log holds 0 entries, so there is nothing to clear.”, built from the same count phrase the Add, Remove and Clear messages use, so every one of them now states how many entries the log holds.

This version also moved the 1.6 entry to the archive page and corrected one word there: the archive said “oldest first” while it lists the newest version at the top, and now says so. Known limits: the wording of that one status message and two lines of archive text — the list, the Show and Order menus, the copied plan, the printed sheet and the day comparison are all unchanged. It was reasoned through by reading the bark log script, not checked in a live browser.

Version 1.22

The hint under the bark log now names the two places a count appears, and says they agree. The Show menu has carried a live count beside each day since version 1.10, and the line under the menus has read “Showing 3 of 6 entries” since version 1.6, but the hint never tied the two together, so the count beside a day (which describes the whole log) and the “Showing” line (which describes what is on screen) could look like they disagree. The hint now says both are counted from the same list: narrowing the view moves the “Showing X of Y entries” line, while the count beside a day keeps describing the whole log.

Known limits: wording on the page only. No script changed, and the counts, the list, the copied plan and the printed sheet are all unchanged.

Version 1.21

The short message under the bark log now says how full the log is after a removal or a clear. Adding an entry already told you the running total, but removing one said only “Entry removed.” and clearing the log said only “Log cleared.”, so you had to count the rows to know what was left. Remove now reads “Entry removed. 3 entries in the log.”, clear reads “Log cleared. Nothing left in the log.”, and after the last entry goes it reads “Entry removed. Nothing left in the log.” The Add message is unchanged.

Known limits: the wording of those status messages only — the list, the Show and Order menus, the copied plan, the printed sheet and the day comparison are all unchanged. It was reasoned through by reading the bark log script, not checked in a live browser.

Older versions (1.21 to 0.1)

Back to the decoder