Data Access Worldwide Forums - DataFlex 2026 Testing

Libraries.................

phvwijk — Sat, 11 Jul 2026 14:14:14 GMT

I have build up over the years a fairly extensive library "construction". Many over the time released in public domain from other dataflex developers. Snippets from the forum, conferences exchanges.
I build extension on and created my own frame work. Well multiple ones.

That is depending on the work I have over time. I worked for more then a few customers doing one to one work. Some of those projects ended up in a one to more customers using it. And I work on a bigger piece of software

All use the my base frame work and can have there own project frame work on top on the base frame work. All good up to now. That worked from Vdf4 up to 2025. For sure some clean up could be done. But why do that if all is working and tested.

But not in 2026.

Compiling since Alpha 2 is not possible. While the same sources compile without a problem in all previous versions.

So the bigger project in the sws file has the following first and major dependency

Code:

{

    "dependencies": [

        "../V-60-Lib/V-60-Lib - 26.0.sws",

Yes there are more but those are not important for the problem.

So that is the main library of the project. That main lib has an dependency

Code:

{

    "dependencies": [

        "../../Vdf12/Libraries/V-26.0/X-Framework/X-Framework.sws",

Again there are more libs there too. But that do not matter

When compiling I get the following error:

Code:

ERROR: 4306 Forward reference not resolved  GET_GIVEDIRECTORYEXISTS ON LINE: 840 (775835) OF FILE: C:\Data\HRp5\V-60\AppSrc\HRp_Head.vw

So double click and go to the offending line. Now use the new Control+Mouse click and it goes to ...... Well it should not go anywhere because I have the compiler error. But it does go to the call

Code:

Function giveDirectoryExists String sDirectory Boolean bShowError Returns Boolean

Ol lets check if the package is part of the view where the error was found. And yes it existing

Code:

Use cDirectoryLogic.pkg // <<<<<==== existing

Ok lets open this file in the studio. And yes it pere4fectly goes the file I showed you the call from.

So what is the problem???

The file cDirectoryLogic.pkg is part of the main lib "V-60-Lib/V-60-Lib - 26.0.sws" of the project.

But there is also a cDirectoryLogic.pkg in base frame work used for all other customers.
The base frame work pkg did not change since 2010 and the main project indeed got an different name for the same call in 2016.

And for what ever reason the compiler finds that first the source in the lower base lib first and the call "does" exists but with a different name. Do not ask me why. If I look in configure Workspace properties at the Library paths

Code:

C:\Data\HRp5\V-60-Lib\AppSrc

C:\Data\Vdf12\Libraries\V-26.0\X-Framework\AppSrc

And the MakePath

Code:

C:\Data\HRp5\V-60-Lib\AppSrc

C:\Data\Vdf12\Libraries\V-26.0\X-Framework\AppSrc

So for me remains the question why the compiler take a pkg file from a to me lower level dependency and the studio work with the to me the proper file.

Because the compiler gives a hick up I did run into this problem. But what if I changed functionality in the same call in a to me higher level and it compiles the lower, that is to me a very bad scenario. To sort that all out I will have to check the prn file for all includes and see which taken first. And check all packages. That is a not so nice scenario.

Is something like I am writing here a weird thing I do, no I do think so. I do it somethings with a pkg from DAW where I make changes in to fix bugs. Sometimes that is done with a subclass. But I prefer to edit DAW with comments what I have done and why. A source compare with a new release alpha, beta, rc and final I will find the differences. With sub classes that is way harder because you forget after a few years you made "upgrades"

But many libs I use have files included made by others. For example Wil van Antwerpen made some sources which are used by others and distributed in the history libs published. It is to me unlikely that all those are getting made ready for the package manager.

Final word. Because the compiler found the coding error I am I think lucky but how can I trust all other stuff what it takes first above what I expect it to do. Will I be the only one effected. I think it is can happen to anyone without knowing it.

Peter

Packages.

phvwijk — Sat, 11 Jul 2026 13:00:41 GMT

How about having a setting/option to include a downloaded package into the pre-compiled part if your application.

A package will be in most likely seen in most cases as something you do not make changes in. so by adding it to the pre compile it saves time in compiling.

Peter

Ppackaging a workspace with multiple .sws files

Ian Smith — Thu, 09 Jul 2026 20:55:54 GMT

With reference to Brams' post (number 402307)

Quote:

we currently want to push an OSS community driven environment.

How do we package a workspace with multiple .sws files e.g.

Capture.PNG

Attached Images

Capture.PNG (16.6 KB)

PKG: Package manager dashboard needs more extensive filtering.

Michael Mullan — Mon, 06 Jul 2026 17:24:21 GMT

I realize the current contents of the pacage manager are "not real" but they do illustrate the need for a better search;

By Author
By Release State, Alpha, Beta, RC, Production
by Release Date/ last change date.
by full text search in the metadata.

and so on.

PKG: Can we have some or all of these workspaces in Package Manager?

Michael Mullan — Mon, 06 Jul 2026 17:18:11 GMT

Code:

43  Graphics Library 2.6    C:\Dataflex\26.0\Fionaidh\libraries\Graphics Library 2.6

   35  StureApsPublicLib       C:\Dataflex\26.0\Fionaidh\libraries\StureApsPublicLib

   21  SigCodejock Classes v1  C:\Dataflex\26.0\Fionaidh\libraries\SigCodejock\SigCodejock Classes v1

   19  VdfQuery                C:\Dataflex\26.0\Fionaidh\libraries\VdfQuery

   14  SigCodejock Wrapper 17.3.0  C:\Dataflex\26.0\Fionaidh\libraries\SigCodejock\SigCodejock Wrapper 17.3.0

   11  cDateTime               C:\Dataflex\26.0\Fionaidh\libraries\cDateTime

   10  Printing                C:\Dataflex\26.0\Fionaidh\libraries\Printing

    5  TinyMceControl          C:\Dataflex\26.0\Fionaidh\libraries\TinyMceControl

    4  AutoComplete            C:\Dataflex\26.0\Fionaidh\libraries\AutoComplete

    2  CodeMaintenance         C:\Dataflex\26.0\Fionaidh\libraries\CodeMaintenance

    2  LibXL                  C:\Dataflex\26.0\Fionaidh\libraries\LibXL 

    1  Chilkat9                C:\Dataflex\26.0\Fionaidh\libraries\Chilkat9

    1  Example Library         C:\Dataflex\26.0\Fionaidh\libraries\Example Library

these are my own, can I publish them?

Code:

 2  DBEKeyPad               C:\Dataflex\26.0\Fionaidh\libraries\DBEKeyPad

    4  ReportTest              C:\Dataflex\26.0\Fionaidh\libraries\ReportTest

    2  Dymo                  C:\Dataflex\26.0\Fionaidh\libraries\Dymo

41  DanesBridgeBase         C:\Dataflex\26.0\Fionaidh\libraries\DanesBridgeBase

Compiler produces a non-deterministic .prn for source containing extended (8-bit / OEM) characters

Nils G. Svedmyr — Mon, 29 Jun 2026 08:13:17 GMT

Hi,

I've run into what looks like a df-cli compiler bug.

When I compile a workspace whose source contains extended / high-bit characters (this one is OEM/CP850, with Swedish � � �), the compiler produces a non-reproducible .prn: compiling byte-identical, unchanged source twice yields two .prn files that differ. The differences are confined to compiler messages that echo a name/symbol containing an extended character.

DataFlex / Studio version : 2026-Beta 1
Compiler used : df-cli command line: "C:\Program Files\DataFlex 26.0\Bin\df-cli.exe" build-file "C:\somepath\myprogram.sre" --workspace "C:\somepath\MyWS.sws" --rebuild --force --write-errors --verbose
Platform : Windows 11 x64
Source encoding : OEM / CP850

What I see:
Two compiles of the same source, no edits in between. The two .prn files are the same length; cmp reports 48 differing bytes, and every one is the same pattern. Example, from a 4538 Obsolete Indicator warning whose text echoes a name containing the extended byte 0x95:

compile A: 4C 95 C5 04 54 95 C5 04 23 50 4F 50 (...OF FILE: L?T?#POP...)
compile B: 4C 95 29 01 54 95 29 01 23 50 4F 50 (...OF FILE: L?T?#POP...)
^^ ===== ^^ =====
0x95 same two trailing bytes differ: C5 04 vs 29 01

The extended character itself (0x95) is identical in both runs. What changes are the two bytes immediately following each extended character � and those bytes are constant within one compile but different between compiles (identical at all 24 occurrences in a given run, then a different value the next run).
That signature � one value reused everywhere within a run, a different value the next run � points to an uninitialized buffer or an address/pointer value written after each extended character (i.e. it tracks process memory layout), rather than a character-set conversion, which would be identical every time.

Why it matters:
Any tooling that parses or diffs the .prn (compile-result analysis, CI baselines, source-modernization tools) sees phantom changes on every recompile because the echoed names aren't stable � and the apparent uninitialized read is worth chasing on its own.
More specifically, it is needed to make reliable analyzis's for the DFRefactor project.

I can supply the two .prn files or a trimmed-down reproducer.

AI generated workspace

Marco — Sun, 21 Jun 2026 19:36:13 GMT

Hi guys

Just mucking around with having an AI generate a workspace by describing the functionality, pointing to webordermobile for sample and .\pkg for packages.

Couple of things that a beginner would not necessarily pickup:

1. Workspace in use (due to IdeSrc) missing - possibly studio could create this folder when missing
2. SQL connect wizard fails as no filelist.cfg in later step of the wizard - possibly studio could detect in wizard and copy from �new� workspace with or without confirmation (and add to workspace path if missing)
3. Webapp* tables missing. - possibly studio could have an option for �add webapp framework tables�

Cheers
Marco

Beta 1: Invalid Message RegisterPublishedProperty

DaveR — Sun, 21 Jun 2026 17:06:25 GMT

Converted WebApp from 25. This not-found procedure seems to belong to cWebObject_Mixin, under 25.0 I had server properties in the Application Object. Do I assume that's now wrong?

B1 latest Toast missing property?

DaveR — Sat, 20 Jun 2026 17:24:16 GMT

Beta 1: Compiling the toastdemo.src seems to have a missing property

Set pbServerOnToastClick to True

I'm still plugging away at setting up rules for conversions from 25 to 26, and though I would replace my 'showtimedmsg' from our windows precompile with something that called toast

Is the toast object allowed to be directly under the application or webapplication objects or must it be in each webobject?

Brealpunt

phvwijk — Sat, 20 Jun 2026 15:02:56 GMT

Although noted in 2025 pretty sure it will be the 2026 version.

It is maybe not something you very quickly experience.

Create a break punt with a Condition.,
Then disable the breakpont
Enable again an your condition is gone.

Peter

Horizontal mouse scrolling

Mike Peat — Fri, 19 Jun 2026 10:22:45 GMT

I have noticed that DF26 (and indeed DF27 ;)), unlike any previous version, supports horizontal mouse/trackpad scrolling. I'm not totally convinced that this new feature is a good thing.

Obviously the ideal is to keep your methods short enough that the whole thing can be seen on one screen (at a comfortable font size), although I confess I'm not at all good at keeping to this. However what I am pretty good at is ensuring that my code very rarely requires horizontal scrolling to be read; I generally keep my lines of code to <= 80 characters wide (yes, I still think I'm using IBM 80-column punch cards; I've been set in my ways longer than most of you have been alive ;)), because I feel that if you can't see a whole line on-screen things get very hard to understand, thus I don't feel any need for horizontal mouse/trackpad scrolling.

In addition, I generally set things up with what Apple call "Natural scrolling" switched off, so when vertical scrolling, a "down" movement takes me "down" the text. In DF26, in the new hozizontal scrolling, a "left" action (on my mouse or trackpad) takes me right in the text and vise versa, which feels (to me anyway) disconcertingly wrong - it is the reverse of what happens in, for instance, Visual Studio Code - in my set up at least, with "Natural scrolling" switched off.

Is there any way of:

Changing the direction of horizontal scrolling
Turning this behaviour off (my preferred option)

Looking through settings on my Mac, my Windows VM and in the Studio, I can't find any way to do these things, but maybe I'm simply missing it. :(

Mike

Beta 1

DaveR — Thu, 18 Jun 2026 13:06:40 GMT

Beta 1: Precompiles must be it the first folder on the AppSrcPath?

Beta1: With 7 folders in the AppSrc list - one error for each entry

Quote:

C:\Program Files\DataFlex 26.0\Bin64\Studio.exe
Referenced Array Index Out of Bounds

Error: 4509

Please report this error to Data Access Corporation

MSG_SAVEWORKSPACEPROPERTIES (17634) - oConfigureWorkspace (1577) - at address 207136
MSG_SAVEWORKSPACEPROPERTIES (17634) - oOK_btn (2000) - in native code
MSG_ONCLICK (4692) - oOK_btn (2000) - at address 206858
MSG_COMMAND (648) - oOK_btn (2000) - in native code
MSG_CREATE_DIALOG (711) - oConfigureWorkspace (1577) - in native code
MSG_CREATE_DIALOG (711) - oConfigureWorkspace (1577) - at address 2217
MSG_ACTIVATE (117) - oConfigureWorkspace (1577) - in native code
MSG_ACTIVATE (117) - oConfigureWorkspace (1577) - at address 3033
MSG_POPUP (290) - oConfigureWorkspace (1577) - in native code
MSG_LOADWORKSPACEPROPERTIES (17633) - oConfigureWorkspace (1577) - at address 207114
GET_EXECUTE@0 (13497) - oStudioCommand (773) - at address 166003
MSG_ONCOMMAND@1 (13516) - oActionManager (655) - at address 105538
MSG_REQUESTCOMMAND (17188) - oConfigureWorkspace (1082) - at address 190716
[start] - at address 291135

SQL drivers

Marco — Wed, 17 Jun 2026 12:31:40 GMT

At FlexRoads something was mentioned about new SQL drivers. I always thought this was the TechStack stream of the product, but some remarks seem to indicate otherwise.

Could this be clarified?

Cheers
Marco

Studio 2026 compiling older code

Marco — Wed, 17 Jun 2026 12:29:26 GMT

Hi, I think I heard at FlexRoads that the new studio can compile older code (as far back as 19.1).
A quick check shows I might have misheard and it�s only the df-cli that can compile older projects.

So did I look not far enough and this could be done?

At several or our customers we have a large varieties of runtime versions. Being able to code any of these in the same studio would be awesome.

Please advise if there is a way

Kind regards
Marco