Sign inRegisterCheck connection

Finding the Right Historical Cumulus Package by Filename

How operators matched Cumulus helper ZIP names and versions—CU-GD-banner versus jpgraph graphs—without treating this host as a download mirror.

Back to Cumulus weather software guides

Forum threads and wiki tables did not say “the graphs thing.” They said a filename and a version: CU-jpgraph-graphs 3.1, CU-GD-banner 3.0, a dayfile parser saved as .php. The dotted URL /projects.downloads on this Cumulus host was how operators matched those names to a job. This article is that catalog of names. It does not attach ZIP files, tracker links, or recovered PHP.

The slash twin, named packages and why the binaries stay offline, explains the rights and provenance reason the files are not here. This page answers a narrower operator question: given a citation, which name did people actually look for? Modern TNET Weather does not distribute these binaries and does not maintain Cumulus.

Historical context

The Cumulus Wiki TNET Cumulus Scripts page still enumerates the family under the names used on the old project page (last checked 13 August 2026). After that page went offline, Saratoga Weather’s legacy scripts continued to list a subset under Saratoga’s own notices, with explicit versions and dates (last checked 13 August 2026). The Wiki marks surviving items with a note that they live there.

Operators found packages by matching strings, not by browsing a generic “downloads” folder. A thread that said “use TNET’s JPGraph pack 3.1” was pointing at a filename. A thread that said “the banner ZIP” was ambiguous until the CU-GD- prefix appeared. Click-tracker URLs and download counters were not the package identity. If the tracker name and the archive name disagreed, the archive name won.

How names were built

Community Cumulus helpers of that period used a small grammar. Reading it still prevents fetching the wrong archive.

Stack prefix. CU- meant Cumulus. Weather Display helpers on the same Saratoga page use WD. A ZIP labeled only “graphs” might be WD sparklines, MX JSON charts, or the 2008 JpGraph pack. The prefix is the first filter.

Job token. GD-banner, GD-avatar, jpgraph-graphs, dayfile named the consumer. Server Check, RealTimeLog, Webtags, WAP, GetMetar, and RSS appeared as Wiki titles even when no Saratoga filename survived. If you cannot map the token to a file contract (realtime.txt, dayfile.txt, web tags, METAR), you do not yet have the right name.

Version and date. Saratoga’s table currently records, among others, CU-dayfile parser 0.12 (12 February 2010), CU-GD-banner 3.0 (20 October 2009), CU-GD-avatar 1.0 (8 October 2008), and CU-jpgraph-graphs 3.1 (26 October 2008). Those numbers are identifiers, not a promise that an unlabeled attachment in a 2014 forum post is the same bytes. A “banner 2.x” cited in an old thread is not banner 3.0.

Extension. Some items shipped as ZIP; the dayfile parser and the avatar script are listed as text to save as .php. Grabbing a ZIP because “all TNET scripts were ZIPs” is how people unpacked the wrong object.

JpGraph as a dependency, not a package name. Saratoga’s CU-jpgraph-graphs row notes that JpGraph must be installed. The graph pack name never included the library version. You still need JpGraph from its vendor under that project’s terms. Confusing jpgraph-4.x.tar.gz with CU-jpgraph-graphs is the classic mix-up this catalog exists to prevent.

Wiki title versus archive name

The Wiki speaks in project titles. Living archives speak in filenames. Operators had to translate.

| Wiki-style title | Filename token operators searched | File contract | |---|---|---| | Cumulus Server Check Script | often “checkserver”; not on the current Saratoga CU list | uploaded realtime.txt or extra file age | | Cumulus JpGraph Graphs Package | CU-jpgraph-graphs | homemade realtime log → PNG | | Cumulus PHP-GD Banner | CU-GD-banner | web tags / realtime → 468×60-class PNG | | Cumulus PHP-GD Avatar | CU-GD-avatar | same packet, smaller canvas | | Cumulus PHP-GD Thermometer | thermometer; not on the current Saratoga CU list | one scalar, analog drawing | | PHP (No Database) Dayfile.txt Parser | CU-dayfile | dayfile.txt → HTML table | | Cumulus RealTimeLog Script | realtimelog; not on the current Saratoga CU list | append snapshots of realtime.txt | | Cumulus TNET Webtags Include Script | webtags include | processed extra file | | Cumulus WAP Dynamic Page | WAP | compact deck from the current packet | | Cumulus RSS Feed | RSS | machine-readable snapshot | | Cumulus GetMetar Include Script | GetMetar | aviation METAR, not a Cumulus internal |

If a name is not in this table, treat it as unverified until a child URL or an independent catalog supports it. Do not invent packages to fill a gap. The named-package essay describes what each surviving job was for; this table is only the lookup.

A later dotted path, /projects.eucookie.pkgs.wdtemp, names a Weather Display–oriented cookie pack. The wdtemp token is the warning: it is not a Cumulus ZIP.

Matching a citation to a flavour

Filename is not enough if the Cumulus flavour is wrong.

Legacy Cumulus 1 versus MX. A 2008 graph pack expects a classic realtime.txt field list (through field 58 in the 1.9.4 / early MX layout). MX later appended feels-like (field 59, v3.7.0) and weekly rain (field 60, v4.3.0) on the realtime.txt map. An old parser that demands exactly 58 tokens will reject a current MX line, or will mis-assign columns if it is sloppy. Record the Cumulus version the filename implied.

PHP version. Saratoga’s page is explicit that some legacy scripts were later touched for PHP 8.x; the CU-dayfile and CU-jpgraph rows are still dated 2008–2010. The name tells you which pack. It does not tell you that it will run.

Library names inside the ZIP. JpGraph’s own archive is jpgraph-4.4.3.tar.gz as of the vendor download page dated 26 January 2026 (last checked 13 August 2026). That is not a TNET name. GD is a PHP extension. CU-GD-banner means the banner script uses GD; it does not mean you install a package called GD from this host.

What operators actually typed

Practical search order, then and now:

  1. Copy the filename token from the Wiki or from the forum post (CU-GD-banner, not “tnet banner”).
  2. Copy the version. If the post has no version, you do not yet have an identity.
  3. Check the stack prefix. Discard WD and unlabeled “graphs.zip”.
  4. Check the extension and the dependency note (JpGraph, GD, no database).
  5. Prefer a living upstream that states a license and a date—currently Saratoga for the four CU items it lists—over an unlabeled attachment.
  6. If the name is on the Wiki list but not on Saratoga, stop. Describe the job from the child article. Do not harvest a random ZIP that happens to contain the word “checkserver”.

Click-tracker URLs, dl.php?id=, and download counters are not names. They rot. The archive filename and version are the durable identifier.

Where to look without using this host as a mirror

  • Cumulus / Cumulus MX: current software from the maintainers, documented on the Cumulus Wiki software page and MX GitHub (last checked 13 August 2026). TNET is not that channel.
  • JpGraph: jpgraph.net, under that project’s terms. See installing JPGraph for include path, fonts, and cache, not for a ZIP of the old pack.
  • Surviving TNET-named Cumulus scripts: Saratoga’s legacy page currently lists four, with a use-at-your-own-risk notice. That is Saratoga’s archive. It is not a TNET mirror and not a restoration of this URL’s old buttons.
  • Rewrite: if you need the function and the grant is missing, write new code against the current file contract.

This hostname’s /projects.downloads URL stays a filename catalog so incoming Wiki and forum links still resolve to the strings they meant.

Modern relevance

Package identity is a data-source problem: which artifact, which version, which stack. A later reader who finds graphs.zip on a backup disk cannot tell whether it plotted Cumulus temperature or WD rain without the filename grammar above. TNET’s data sources, quality controls, and methodology note is the modern place that kind of claim is treated for public records used in connection research. This catalog does not attach those records, and that research page does not ship PHP.

The projects map groups the same names as engineering problems. The Cumulus legacy hub lists related pages.

Sources