Who is familiar with translating AND the test functions?

[tgtje]

I stumble into an issue whilest using a new LAN define.
My translation for History (history.php) is 90% done and runs fine.
But theres is an issue where i want to translate (text) : Table

` 'history_table' => array ( 'title' => 'Table', 'type' => 'text', 'data' => 'safestr', 'width' => 'auto', 'filter' => true, 'help' => '', 'readParms' => [], 'writeParms' => [], 'class' => 'left', 'thclass' => 'left',),

   'history_table'           => array ( 'title' => LAN_HISTORY_TABLE, 'type' => 'text', 'data' => 'safestr', 'width' => 'auto', 'filter' => true, 'help' => '', 'readParms' => [], 'writeParms' => [], 'class' => 'left', 'thclass' => 'left',),`

My install itself works fine (English and Dutch) NO error report etc. (php 8.5.6) latest master

However when i want to make the PR (regardless iff it will be accepted), the test on Github (which runs after uploading into a fork).
keeps failing on this LAN alone.
UNDEFINED CONSTANT
all others (10 other new LAN's EXcluding this one) have no issues at all (LAN's into lan_admin as no own adm/lan_hist...php exist).
Anyone a pointer for me ?

note
( unfamiliar terrain : admin_ui.php is the piece where it happens/clashes but forgot the exact line numbers (dumped the fork) (between L 1522 and 1665)
here's a look of system and runs as expected. ps taken another admin theme do display that all column naes have a 'underlink'(sort asc/desc) which is not always visable in another admin theme.

Thanks
ps this is MASTER only (no 2.311 and earlier

	`

[e107help[bot]]

Nothing in the tests is against you here, Leo, and the proof was already sitting in your own fork.

The one thing to do, when you rebuild the branch: give the constant the same name in both files, and judge only the run for a commit where both files carry it. Which name is up to you. With LAN_HISTORY_TABLE:

// e107_admin/history.php
'history_table' => array ( 'title' => LAN_HISTORY_TABLE, 'type' => 'text', ...

// e107_languages/English/admin/lan_admin.php
'LAN_HISTORY_TABLE' => "Table",

Why it kept failing on that one. I read the eight Unit Tests runs on your fork from 1 September (times UTC). You renamed the Table constant three times, and the web editor commits one file at a time, so every rename produced one run on the half-done state and one on the finished state:

Commit	history.php asks for	lan_admin.php defines	Unit Tests
384807dd82 13:23	LAN_HISTORY_TABLE	nothing yet	red
065ab538e3 13:24	LAN_HISTORY_TABLE	LAN_HISTORY_TABLE	green
838327163c 14:05	LAN_HISTORY_TABL	LAN_HISTORY_TABLE	red
4cfc749efe 14:06	LAN_HISTORY_TABL	LAN_HISTORY_TABL	green
74457c7d83 14:34	LAN_HISTORY_TABL	LAN_LABLE_HISTORY	red
3d98cc34d3 14:35	LAN_LABLE_HISTORY	LAN_LABLE_HISTORY	green

Every red is a commit where the two files disagree. Every green is one where they agree, and those greens ran the whole suite with all eleven constants in place. After 065ab538e3 the other ten never changed name again, so they never failed again. An install where the two files agree, like yours, has nothing to complain about.

Why the first red named only one constant. At 384807dd82 all four column titles were missing from lan_admin.php, not just Table. PHP builds the $fields list top to bottom and stops at the first name it cannot find, so it reported LAN_HISTORY_TABLE and never reached LAN_HISTORY_ACTION. That is where "this LAN alone" came from.

Reading it next time. The test is scriptsTest, which loads every admin script the way a browser would and fails on anything PHP objects to. In a red Unit Tests job, this is the block to find:

Scripts that did not load cleanly:

history.php: exit 255 after 0.16s
  Fatal error: Uncaught Error: Undefined constant "LAN_HISTORY_TABL" in /var/www/html/e107_handlers/admin_ui.php:1661

The name in quotes is the one to search for in lan_admin.php. admin_ui.php is only where PHP first builds the class; the line asking for the name is line 78 of history.php, the one that still reads 'Table' on master.

Two things for the PR:

• init() still has a hard-coded Restore label, right under LAN_DELETE and LAN_UPDATE.
• Deltik asks that a new constant say what it is rather than carry a number, so something like LAN_HISTORY_HELP_INTRO in place of LAN_HISTORY_HELP_01. LAN_HISTORY_TABLE and LAN_HISTORY_ACTION are the shape he wants.

A red run describes the one commit it ran on, not your install. The check is whether both files carry the same name at that commit.

[tgtje]

Thanks for the info (Yes tested: found mistake repaired etc.retests run etc..), i will a run new test again (after all these years, i still learn) and surely make mistakes.
One side note LAN shape : something like LAN_HISTORY_HELP_INTRO
i do not have problems with that, BUT (yes there's a but).
The help on left side of admin page is build in a html form, and reads different lines (as paragraphs p; and strong (for emphasize subtitle)) in < li>...(list)lines.
Hence those occupy (in a working state (might be not the proper way)) 6 lines (6 LAN's) eg number. (OR i could use a word out of the line)
even if it is a rule or recommended not to use html in define texts i can not achieve the emphasize without it ( used <b ; (bbcode no go) instead of strong), for these. Would i use H4 or alike > html..., > searched smallest impact; it works for a translation (visual) ;-)

Restore 'text' (init line) will be taken in.

(why did i ask now : i want/would like to make a larger PR with ALL that is not yet in translatable form.; so there's only 1 big instead of (10 in total smaller ones.)
for those interested : changes in banner; chatbox; newspost ; social; admin prefs (cinfo); featurebox; forum (admin); admin logs; history *; blank plugin.......... most are ready

[tgtje]

ps forget the 1 large PR; i forgot BC compatible. Have to take in account define version(2.3..);
History likely the only (not shure) for master (have to test others)

[e107help[bot]]

Good to hear the retest is on its way, Leo. Three answers, shortest first.

Names for the six help lines: take a word from each line. Six lines are six constants either way; the name only changes what someone reading history.php sees. Your text stays as it is, only the names change:

// e107_languages/English/admin/lan_admin.php
'LAN_HISTORY_HELP_INTRO'    => "This page allows you to view the <strong>history of changes</strong> made to records in the system and restore records to a previous state when needed.",
'LAN_HISTORY_HELP_FEATURES' => "Features of this page:",
'LAN_HISTORY_HELP_VIEW'     => "<strong>View Changes:</strong> See details of updates and deletions, including who made the changes and when.",
'LAN_HISTORY_HELP_REVERT'   => "<strong>Revert Changes:</strong> Restore a record to its earlier version, undoing accidental or undesired modifications.",
'LAN_HISTORY_HELP_AUDIT'    => "<strong>Audit Trail:</strong> Track all actions performed on records for accountability and transparency.",
'LAN_HISTORY_HELP_FILTERS'  => "Use the filters to narrow down the history logs or locate specific changes. If a record can be restored, an option will be available in the Options menu.",

and the same six names in renderHelp() in history.php, where your _01 to _06 are now.

A tag inside the string is fine, and you can keep the one the file already has. Core ships it: line 97 of lan_users_extended.php has <b><i>Parameters:</i></b> inside a help constant. The lines you are converting say <strong> today, so there was no need to swap to <b>. Your split is the right one: the <p>, <h4> and <li> stay in history.php, only the words go to the language file, and a <strong> around two of those words travels with the words. Bbcode did nothing for you because history.php hands the help text straight to the page. The cron page runs its text through toHTML() first, which is why its constant can use [br]. Both routes are shipped. Yours is the smaller change.

Master and 2.3.x. Of your list, history.php is the only page that does not exist on release/v2.3.x; the rest are on both, banner included. The language files differ in form: master's are return [ arrays, 2.3.x's are define() lines, so a master PR cannot be copied across as it is.

So the history PR has no twin at all. For a page that is on both branches, the master PR is the one to open first, and the define() version for release/v2.3.x is a second PR. You have written more of those than anyone here. If a particular twin is one you would rather not write, ask on that PR once it exists; whether I open it is Deltik's call, as it was on #6134.

One PR per page, my own view rather than a rule: the sweep reports each script on its own, so a red on one page does not block the others.

One rule that does matter across ten files: add constants, never rename or delete an existing one, whatever it is called. Other language packs key on those names, your Dutch one included.

History first, and the rest one page at a time.