The screenshots
Every screenshot on this page is a photograph of a page that runs. The archived source
was converted to something a current web server can execute, served locally, and rendered
in a browser at 1280 pixels wide. Nothing is a mockup and nothing is redrawn.
Every row of data behind them is invented. The original databases hold a real shop's
real stock, so the tool reads them only to learn their shape, and then writes new rows into
it: the same tables, the same number of rows, none of the same values. The headings, the
labels and the layout are the real ones. The wheels are not.
How each date is known
- 2012, the shop's site. Nothing on the page dates itself. The archive dates
the files in that tree from 4 January 2010 to 19 August 2012.
- 20 January 2012, the catalog. The file's own date in the archive.
- 25 May 2012, the inventory. The database table this screen reads carries a
timestamp in its own name, which decodes to that day. The file holding it was last
written on 4 June 2012.
- 31 October 2012 and 31 January 2013. Mine, and not in the archive. Those two
dates come from me.
- 2014 and 2015, the rim sheets. The file names carry their own dates. Eight of
them survive, running from 28 February 2014 to 6 August 2015.
- 2019 to 2020, the classic build. Thirty seven item number exports carry dates
in their names, from 28 May 2019 to 11 August 2020, and the upload folder carries sheets
from 2019 and 2020.
- Early 2026, the rebuild. The classic build's own notes say so, and they
say it approximately, so this page does too.
The eleven columns
Three rim sheets survive with three different column sets. Folding case and spaces,
eleven names appear on all three: Color, Pics, Size, Bolt, OFST, Holl, FR, ItemNumber,
PartNumber, Price and Notes. Four names from the February 2014 sheet are gone by April:
Vehicle, Years, TPMPVS and Imper, each of them replaced by something finer.
That is a measurement of these three files and nothing more. It is not the same eleven
as the eleven of thirteen in the story above, which is about the hand kept spreadsheet and
the live table today, and which cannot be checked here because the spreadsheet is
missing.
What is not here, and why
The spreadsheet has not been recovered. It is not in the archive, not in the backups,
and not on any disk that has been searched. If it turns up it goes in the frame that is
holding its place.
The first Templatizer is not an exhibit. Its code survives in full and parses clean,
but there is no database of any kind in that tree, of any format, so every page of it dies
at the first query. Its data does survive, separately, as a SQL Server backup, and when
that is restored and proved to hold nobody's details, the tool becomes an exhibit like
the other two. It is not shown running on a database borrowed from a later version.
The current system is not photographed. It is a live business with real orders in it,
and the only honest way to show it is a copy running on invented data.
Two pieces of the classic build are left out of its exhibit and say so at their own
addresses: a second application for the shop's non factory stock, which has its own
database and its own mailing list, and one exporter that reaches out of the archive into a
system that is still running.
And one page of the classic build is simply destroyed. The center cap listing board is
4,649 bytes of noise in the archive, with no readable text in it and no header saying what
it used to be, and it is the only damaged file among the 154 there. A file of the same name
and exactly the same size survives in a different copy of the tool, but that copy differs
from this one in every other file compared, so it is not laid in. There is a page at the
address saying what happened, and the rest of the center caps module runs.