Sprezzatura :: Making Databases Happen

By Sprezz | Wednesday, 1 March 2023 17:42 | 0 Comments

We recently came across a situation that we have literally never encountered in our decades of working with Linear Hash. Basically a FIXLH operation was interrupted mid process and hi-jinks ensued. The resultant mess took a while to sort out but our findings may prove useful if you ever find yourself in a similar situation.

This specific issue was encountered on an AREV 3.12 system but the logic for FIXLH remains roughly the same so the information remains applicable.

Basically, the FIXLH operation was interrupted about 10% of the way through processing (the NTVDM abended). A quick check of the table header revealed that it believed itself to be a clean empty table with no rows. Yet listing the table produced a set of row ids after an initial pause while nothing was apparently happening.

What had happened was that the table header had been reset to that of an empty table, and the first 10% or so of the primary frames had been blanked. This caused us to investigate the processing undertaken by FIXLH and we can confirm that it is as follows:

Create and/or clear the temporary dumpfix table

grab frame zero group

write rows in group to temporary dumpfix table

set frame zero header to that of an empty table

loop

grab next group

write rows in group to temporary dumpfix table

wipe primary frame to null

until no groups left

repeat

copy rows in temporary table back to source table

So if the operation is interrupted, some of the data will be in the source table and some will be in the temporary dump fix table.

As an additional caveat, note that if the rows in the temporary dump fix table are not stored off elsewhere, they will be lost next time FIXLH is run.

Since the advent of the UD we've never had to actually fix a GFE for quite some time, so this was an eye-opening operation!

Backups are your friend...

Sort of obvious in hindsight

By APK | Tuesday, 31 January 2023 16:04 | 0 Comments

Recently at a client site, we encountered an error that was a little more baffling than normal. Under very specific circumstances,    RLIST would bail with no results. Eventually, we had a duplicatable test sentence which was basically

   SELECT TABLE BY CALC_COL_1 WITH REAL_FIELD_1 = ‘XXX’ AND WITH CALC_COL_2 = ‘YYY’

When we removed the BY clause (CALC_COL_1) our sentence ran to completion.   This put the problem squarely on CALC_COL_1, which is where we started debugging.

We changed CALC_COL_1 to a simple “@ANS=1” and ran the sentence again. This worked correctly. To prove to ourselves that CALC_COL_1 was the problem, we put back the original code and changed our sentence to

    SELECT TABLE BY CALC_COL_1 WITH REAL_FIELD_1 = ‘XXX’ AND WITH CALC_COL_3= ‘ZZZ’

This also worked.

Now we’re starting to get a little confused. Modifying CALC_COL_1 solved the problem. Replacing CALC_COL_2 also solved the problem. It has to be some sort of interaction with the calculated columns, because there’s no obvious consistency lining up the errors with the sentence structure.

It’s time to break out the profile log and see if something jumps out, and it did. Something appeared off with EXTERNAL.SORT. It seems like it executed a few times, and then everything stopped. We dusted off our copy of PROCMON.EXE¹ and ran the process again. PROCMON showed that the sort file was being opened, but that it was not being closed. It also showed a lot of file access We’re now convinced this has something to do with sorting in general, and not with the specific field being sorted.

After trying well over a dozen different variations on the RLIST sentences, we finally worked out that we needed two things to happen for the error to occur. The first was that we had to have CALC_COL_1 as the sort column. The second was that the WITH clauses needed to return enough keys to force the system into using EXTERNAL.SORT instead of a simple in-memory sort.

Normally, when OpenInsight starts behaving very oddly, the culprit is usually a function that’s not clearing set_status() before calling get_status. When that happens, the get_status() call could be returned an error from 10 programs back. We reworked the calculated columns to ensure we clear set_status, and feeling very proud of ourselves, sat back and waited to bask in the glory offered to us by our client. Obviously, the report failed again.

Sitting back, we start wondering if maybe there’s a small part of the RLIST code that’s still thinks it’s Advanced Revelation and it’s actually checking the pseudo STATUS() variable. So, we reworked the calculated columns to ensure we clear set_status and status(), and feeling very proud of ourselves, sat back and waited to bask in the glory offered to us by our client. Obviously, the report failed yet again.

Surely there can’t be three ways to set an error condition in OpenInsight? It turns out that there is. We suddenly remembered @FILE.ERROR. So, we rework the calculated columns to ensure we clear set_status and status() and @FILE.ERROR, and feeling very proud of ourselves, sat back and waited to bask in the glory offered to us by our client. And this time there was much rejoicing and basking.

With the solution in hand, we set out to find out exactly why this happened and where the system was registering the failure. What we found was the calculated field was doing a lot of file access and would attempt to access records that didn’t exist. This was what was setting @FILE.ERROR. But, we hit a two-fer, as they say. Some of the randomness had to do with the amount of data returned. If enough sort data was returned the system would call EXTERNAL.SORT to sort the data. EXTERNAL.SORT writes to temporary OS files, and the OS file commands use @FILE.ERROR to check for errors. EXTERNAL.SORT would execute an OS operation, check @FILE.ERROR, find the pre-existing error, and abort the process. But, this would only happen if the very last row processed set @FILE.ERROR. Otherwise RTP57 would clear any prior @FILE.ERROR settings, and @FILE.ERROR would be null during the EXTERNAL.SORT call.

1. PROCMON and other useful tools are available as part of the Sysinternals suite of development tools, which no programmer should be without.

Y2K Redux - Indexing Issues With Dates OR 20100 - A Search Oddity

By Sprezz | Thursday, 12 January 2023 12:46 | 12 Comments

Those of you who've been in the industry for a while will remember the smugness with which we treated the Y2K issues the rest of the world were experiencing. It didn't affect us, we stored our dates in internal numerical format. I'm sure we all have favourite anecdotes. Mine was being called into the Press Office of a major Government department to certificate their AREV tracking system. I sat down and gave the system a cursory glance. "Ermmm - it doesn't seem to use dates?". "No, it doesn't". "Well, here's your certificate"... turns out they couldn't get the budget for system tweaks but they could for Y2K compliance. I spent the rest of the day implementing their desired changes.

Anyway, this week wiped the smile off of our collective faces (if you'd ignored the advice in KB 42 thirty or so years ago - which to be honest you could be forgiven for) with the advent of our very own variant - we'll call it the 20100 bug, although it's not a bug, it's an unfortunate feature.

Users began reporting that date searches were failing for values after the 10th of January 2023. At first we couldn't see an obvious reason. We built a database containing date values going back to the previous century with five rows per date and indexed it. We wrote a test program and ran it

RUN ZZ_POP 10/01/2023

RUN ZZ_POP 11/01/2023

RUN ZZ_POP 12/02/2023

But then a throwaway remark from Martyn about pivot dates being discussed in the MV community led to a realisation. What if we used internal dates?

RUN ZZ_POP 20099

RUN ZZ_POP 20100

BINGO!

What we've fallen foul of is discussed in the KB reference earlier. Basically BTREE.EXTRACT takes what it is given and tries to ICONV it to use for the look up. If it can't ICONV it (20099 isn't a viable date) it uses the value passed. If it CAN ICONV it, is uses the ICONVed value. And guess what? 20100 ICONVs to the 2nd of January 2000 as you can see above.

Of course, this won't be the first time this has happened in the wild - for example looking for the 7th of December 1998 using 11299 would have returned the 1st of December 1999 and so on.

Note that the same issue will be experienced when using internal date formats with RLIST statements.

The solution? When calling BTREE.EXTRACT use EXTERNAL values as it will try and ICONV the data before using it.

Addendum

We've been asked to share the code we used to locate btree.extracts in client systems to enable a manual inspection to determine possible failure points. This is a rough and dirty hack which met our requirements. Feel free to tailor to your own requirements.

0001  Subroutine zz_find_Btree_Extract( void )
0002  /*
0003     Author      AMcA
0004     Date        Jan 2022
0005     Purpose     Quick hack to help identify system usage of btree.extract
0006                 Provided as is with no warranty
0007  */
0008    
0009     Gosub initialise
0010     Gosub process
0011     
0012  Return
0013  
0014  initialise:
0015  
0016     filesToCheck   = "SYSPROCS,SYSREPOSEVENTS,SYSTABLES"
0017     columnsToCheck = ",,8"
0018  
0019     loopCtr        = Dcount(filesToCheck, ",")
0020     resultString   = ""
0021     
0022  Return
0023  
0024  process:
0025  
0026     For loopPtr    = 1 To loopCtr
0027        file        = Field( filesToCheck, ",", loopPtr )
0028        column      = Field( columnsToCheck, ",", loopPtr )
0029  
0030        resultString := file : \0D0A\
0031  
0032        if file = "SYSTABLES" Then
0033           starting = "%"
0034           Gosub processSysTables
0035        End Else
0036           starting = "@"
0037           Gosub processRest
0038        End
0039     
0040        resultString := \0D0A0D0A\
0041  
0042     Next
0043  
0044     Oswrite resultString On "ZZ_BE.TXT"
0045     Call Set_Property("CLIPBOARD", "TEXT", resultString )
0046     
0047  Return
0048  
0049  processRest:
0050     Open file To v Then
0051        Gosub processV
0052     End Else
0053        Call FSMsg()
0054     End
0055  Return
0056  
0057  processV:
0058  
0059     Select v
0060     eof = 0
0061     Loop
0062        Readnext id Else eof = 1
0063     Until eof Do
0064        If id[1, 1] = starting else
0065           Read row From v, id Then
0066              ptr = 1
0067              If column Then
0068                 saveRow = row
0069                 row = row< column >
0070                 Convert @Vm To @Fm In row
0071              end
0072              lenRow = Len(row)
0073              lineNo = 0
0074              If lenRow > 0 then
0075                 Loop
0076                    nextline = row[ptr, @Fm]
0077                    ptr = Col2() + 1
0078                    lineno += 1
0079                    there = IndexC( nextLine, "btree.extract(", 1)
0080                    If there Then
0081                       variable = Trim( nextLine[ there + 14, ","] )
0082                       If variable[1, 1] = "'" Or variable[1, 1] = '"' Then
0083                          // passing literals - check the line
0084                          resultString := file : \09\ : id : \09\ : lineNo : \09\ : nextLine : \0D0A\
0085                       End else
0086                          // ok we're now going to work back through the code until we find = for our var
0087                          tempLineNo = lineNo
0088                          Loop
0089                             tempLineNo -= 1
0090                             line = row< tempLineNo >
0091                             If Trimf( line )[1, 1] = "*" Then
0092                                line = ""
0093                             End
0094                             Convert " " To "" In line
0095                             line  = " " : line
0096                          Until IndexC( line, " " : variable : "=", 1)
0097                          Until tempLineNo = 0
0098                          Repeat
0099                          If tempLineNo = 0 then
0100                             resultString := file : \09\ : id : \09\ : lineNo : \09\ : "Not found " : row< LineNo > : \0D0A\
0101                          End else
0102                             resultString := file: \09\ : id : \09\ : lineNo : \09\ : row< tempLineNo > : \0D0A\
0103                          end
0104                       end
0105                    End
0106                 While ptr < lenRow
0107                   
0108                 Repeat
0109              End
0110           End Else
0111           end
0112           Call send_info( file : " " : id )
0113        end
0114     repeat
0115  return
0116  
0117  processSystables:
0118  
0119     Open file To v Then
0120        Select v
0121        saveV = v
0122        eof = 0
0123        dictCtr = 0 
0124        
0125        Loop
0126           Readnext id Else eof = 1
0127        Until eof Do
0128           If ID[1, 5] = "DICT." Then
0129              Open id To v Then
0130   
0131                 file = id
0132                 
0133                 Call push.Select( v1, v2, v3, v4 )
0134                 Gosub processV
0135                 Call pop.Select( v1, v2, v3, v4 )
0136                 eof = 0
0137  
0138              End Else
0139              *   Call FSMsg()
0140              End
0141              v = saveV
0142           end
0143        Repeat
0144     End Else
0145        Call FSMsg()
0146     End
0147  
0148  return

Owned - By A Panel

By Sprezz | Wednesday, 4 January 2023 10:16 | 0 Comments

Tabbed interfaces provide a much loved way of fitting lots of information onto a single entry form. The introduction of multi page forms and the improved integration of page controlling made this a much easier exercise - each tab could be a page of the form and we could simply move between pages on a tab click. In fact a quick event was added to make this even easier.

But what has not been possible until now, without a deal of smoke and mirrors, is to have one or more tab controls embedded within a page on a single or multi-page form and have them only appear on a specific page. The only way to accomplish this would be to have a set of controls per tab, all within the tab boundaries, and make them visible or invisible depending on the currently clicked tab. This makes form maintenance incredibly tedious and code overly complex.

Before OI10 there was no real concept of parent/child relationships in OI - other than that forms and group boxes owned all of the controls within their boundaries. This led to many issues, not least with group boxes and tab controls. With the advent of OI10 this has changed. OI now allows the implementation of, and respects, parent/child relationships.

The biggest example of this is the introduction of a "Panel" container type.

We are going to use the "Panel" (not the "Simple Panel") to enable tabs within a form. But first, what's the difference between a "Panel" and a Simple Panel?

A Panel is simply a container that can be placed anywhere on the screen so that when controls are pasted into the Panel they become children of the Panel, not of the Window. So move the Panel and the controls owned by the Panel move with it. Don't be too put out by the "Simple" part of the description, pretty much the only difference between a Panel and a Simple Panel has to do with paging. So looking at Simple Panel/Panel properties in the Form Designer we see this :

and comparing the two sets of properties we see that the only properties unique to Panels are

PageCount
   Virtual Size
   PageSwapRenderMode
   ScrollMode
   TabWithinPanel

and the only property unique to Simple Panels is the DummyCaption property.

So, to implement a tab within a form page, we can now combine the Tab Control, the Panel Control and one simple Quick Event to achieve the desired results.

Let's start with a simple one page entry screen.

Select "Panel" from the Containers and position it within the tab control - using Right-Click and drag to size it appropriately. (See Addendum for other keystroke combinations). There are no hard and fast rules about this but I tend to leave an 8 pixel border from the tab control to make it easier to edit in the form designer.

(As an aside, if you right-click on the Window you can un-anchor it to allow moving it wherever you want.)

Tell the Panel that it has three pages and label the tab accordingly (after setting the tab to three values) :

Select the Panel and add in the controls for the first page.

Now select the Panel and use the spin control to move to page 2 and add in the controls for the second page. Note that the spin control applies to whichever control is both currently selected and supports paging - namely Windows and Panels.

And finally the third page.

Finally we need to add a Quick Event on the Tab click. This needs to get the VALUE Property of the Tab control, and set the CURRENTPAGE Property of the Panel.

and Bob, as they say, is your uncle...

Actually a quick revisit - internal discussion at Sprezz Towers brought up the fact that the article missed a fairly big point (because the author ass-u-med it was obvious which it obviously isn't :)). Panels can host panels that can host panels - as far as you are practically likely to go. So in the screen below we see multiple panels in action.

Addendum - Mouse Use In Form Designer

When in Draw Mode:

Left Click

1) Left-click to drop a default-size control on a the form

2) Left-click and drag to draw a control on the form

3) Ctrl-left-click to drop a default size control on a container control

4) Ctrl-left-click and drag to draw a control on a container control.

Right Click

1) Right-click and drag to draw a control on the form

2) Right-click and drag to draw a control on a container control.

When in Normal Mode:

Left Click

==========

1) Left-click on the form to unselect all controls

1) Left-click on an unselected control to select it

2) Left-click and drag around/across a group of unselected controls on the form to select them

3) Left-click and drag on an unselected or selected control to move it

4) Left-click and drag around a group of selected controls to move them

5) Ctrl-left-click on a selected container control and drag around/across unselected child controls to select them

6) Ctrl-left-click on a selected container control and drag around/across unselected child controls to select them

7) Shift-left-click to add an unselected control to a group of selected controls with the same parent

8) Shift-left-click on a group-selected control to remove it from a group of selected controls

Right Click

===========

1) Right-click on the form to display it's context menu

2) Right-click on an unselected control to select it and display it's context menu

3) Right-click on a selected control to display it's context menu

4) Right-click on a group of selected controls to display the group-selection context menu

5) Right-click and drag on the form to draw the most recently added type.

6) Right-click and drag on an selected container control to draw the most recently added type and add it as a child.

7) Right-click and drag on an unselected container to draw the most recently added type and add it as a child.

8) Shift-right-click on the form to drop the the most recently added type with a default size.

9) Shift-right-click on an selected container control to drop the the most recently added type with a default size and add it as a child.

10) Shift-right-click on an unselected container control to drop the the most recently added type with a default size and add it as a child.

You are Auth-ful - but I like you

By Sprezz | Thursday, 24 November 2022 14:42 | 0 Comments

Recently, a client was a little late in applying their authorisation code to their copy of OI10, and attempts to access OI were met with error messages "PSSI.66: ???". This seemed a strange message and at first we suspected that it wasn't even an OI message.

However chatting to the Rev developers, it seems that an insert (PS_RESOURCES) was missing from SYSENV and that it should in fact have had the following values

///////////////////////////////////////////////////////////////////////////////

//[PS_SYSINTIALIZE]////////////////////////////////////////////////////////////

PSSI.00: No entry window defined for the %1% application.
PSSI.01: No visible entry window started for the %1% application
PSSI.02: Unable to start entry window "%1%" for the %2% application
PSSI.20: . Process aborted
PSSI.21: %1% initialization
PSSI.60: This Developer Class Server Deployment Pack for OpenInsight is due to expire on %1%.|Please contact Revelation Software at +1-800-262-4747 to renew your license.
PSSI.61: This copy of OpenInsight is due to expire on %1%.|Please contact Revelation Software at +1-800-262-4747 to renew your license.
PSSI.62: This copy of OpenInsight expired on %1%.|Use of this software is in violation of your OpenInsight Software License Agreement.|Please contact Revelation Software at +1-800-262-4747 to renew your license.
PSSI.63: This copy of OpenInsight has expired.||To purchase this product, contact Revelation Software on the web at:||www.revelation.com||Or call toll-free: (800) 262-4747
PSSI.64: This evaluation copy of OpenInsight expires on %1%.||To purchase this product, contact Revelation Software on the web at:||www.revelation.com||Or call toll-free: (800) 262-4747
PSSI.65: This evaluation copy of OpenInsight has expired.||To purchase this product, contact Revelation Software on the web at:||www.revelation.com||Or call toll-free: (800) 262-4747
PSSI.66: This Developer Class Server Deployment Pack for OpenInsight expired on %1%.|Use of this software is in violation of your OpenInsight Software License Agreement.|Please contact Revelation Software at +1-800-262-4747 to renew your license.
PSSI.67: Unable to process License Information.
PSSI.68: OpenInsight Licensing

So that clarified that issue.

The issue that we were left with was that normally in OI10, reauthorisation is done from within the product - and if you can't log in... you can connect the dots.

So, we went on search for a reauthorisation executable in the OI directory and not surprisingly found revauth.exe. This is a small executable, and clicking it seemed to do nothing. Until we realised that it was likely a command line executable.

Invoking powershell, we moved to the UNC containing revauth.exe and executed it with a /?. These were the results.

PS Microsoft.PowerShell.Core\FileSystem::\\MyServer\Shares\Sprezz\programs\OINSIGHT_10> .\revauth.exe /?

revauth <featurename> <authcode> <filename>

values for <featurename>:

OI : OpenInsight
UDN: Universal Driver (network user license)
UDH: Universal Driver Heavy

So we instructed the client to run the following command

revauth OI AuthCode revengine.lic

and lo and behold, the system worked again!

As an aside, the licensing information is no longer stored encrypted in a DLL but rather is stored in the revengine.lic file where it can be easily viewed - a modified example is shown below.

Of course, you can't alter any of these values in notepad as they are actually encoded into the signature, but nice try!

And for readers who aren't British and of a certain age...

OI 10 makes it easier to get onto the Property ladder

By Sprezz | Saturday, 1 October 2022 21:02 | 0 Comments

One of the intermittent problems we face when developing with OI 10 is knowing about the new properties and features that we now have access to. The pace of development is so rapid that the documentation just doesn't keep up. There's a fine source of documentation at revDevX.com along with multiple blog posts at the same site but sometimes the Lord helps those who help themselves.

With this in mind, we'll use a real world example to illustrate how much easier life can be when we have access to such new features.

We recently wanted to have a radio button contract or expand based upon a control record. We knew how we would have accomplished this in the past - we'd grab the button control object, destroy it, modify it then recreate it. That's all well and good, it works and the control stays at the right place in the tab order - but it can change the Z-Order of the control which isn't always desirable.

Naturally we wondered if there was a better way in 10.

Of course there is, or we wouldn't be making this blog post.

Taking a simple example where we want to alter the available buttons based on the membership type, we'd have one set of radio buttons for the member type and another for the letter type.

An existing member wouldn't need a Welcome Letter so clicking on "Existing Member" has to do this

i.e. the "Welcome Letter" option is removed.

We all like to boast that OpenInsight is written in OpenInsight but the implications of this can be missed in the excitement of the new environment. We're so used to OI not exposing all of a control's properties that we lose track of the possibility that what we need may be staring us right in the face.

The simple fact is, that if something exists in the property panel in the form designer for a control, then it is an exposed property. So let's look at the property panel for a radio button.

and what do we see as the fourth property? Buttons.

Clicking on the options button displays a dialog containing the text and values for the radio button :-

As we've telegraphed this far in advance you can probably guess what this means - radio buttons expose a BUTTONS property. It consists of two columns, both multivalued. The first is the label description and the second is the associated code.

So now, to accomplish the above goal all we have to do is manipulate this property. Here's a simple code snippet to do this :-

If you're not familiar with the syntax being used for getting and setting properties check out the blog post here.

We'd encourage you to take a look around the property panels to see what other goodies you can find!

Indexing issues on LARGE tables

By Sprezz | Tuesday, 7 June 2022 14:25 | 0 Comments

We've been doing a lot of stress testing of LH at Sprezz Towers to help a client with speed issues on large tables. As part of that we've created some 60 million row tables to play with combinations and permutations of differing indexing setups.

In so doing, we came across a situation where the index sort would get to 80% and then just exit with an FS220 error. (FS_REL_NEED_REBUILD_ERR$). Now to be honest, this was confusing to us as there wasn't a relational index on the table. So speaking to the good folks at Rev, it was clarified that this was the wrong error but for fairly obvious reasons we were the first to notice. After discussing further with them we came to the conclusion that it must be related to disk space.

So began an increasingly frustrating attempt to free up disk space (why oh why did I format my USB sticks with FAT32 and its 4GB file size limit?). Subdirectories were zipped, old files deleted until - finally we had the 40GB or so free that we needed.

We tried again - and again around the 80% mark the index rebuild just stopped. It didn't crash, it just stopped - having cleared out the %ALL.IDS% token in the ! file.

As V119 was suspected to be the culprit, we redirected the temporary sort file in the environment to the same disk as the data tables so that we could keep an eye on what happened.

At 77% and the sortfile is growing

78% and still going

and then - boom

It seems that the largest sort file that can be created is 2GB and that attempts to exceed this will fail. Note that this applies to both IDX_SETS1 and 2. This is a limitation of a 32 bit OS rather than OpenInsight, so can not easily be addressed in version 9.x.

Now it should be pointed out that this is a particularly extreme situation, as the row ids in this large table are 5 part and quite long. With a more normal key structure than the long complex one we were using, it is unlikely that this limit would be breached without having hundreds of millions of rows in the table... but it does mean that if you need to rebuild such large indexes you need to undertake the task in OI 10.1 where the issue has been resolved. Note also that this doesn't affect day to day use of the indexed table, normal additions and deletions will still update the index - this issue only affects a rebuild.

As a side note - if you DO want to do this in 10.x there is a major caveat. Indexing has been rewritten for 10 and works in a different way than 9. This means that out of the box, OI 9 and OI 10 have different indexing routines and are not compatible. Fortunately Rev have provided a way to deal with this.

All that we need to do is edit or create a record in SYSENV called CFG_RTI_UPDATE_INDEX. Set line 1 to RTI_UPDATE_INDEX_90 then save it, exit OI and restart. This will force 10 to use 9 indexing logic. We've been working a lot with indexes on large tables in 10 and we'd recommend using this setting in any case if you're working with large data volumes. By large tables, we are talking tens of millions of rows.

For the avoidance of doubt - if you are sharing data between OI 9 and OI 10 you MUST do this or you will experience issues.