Mail Me

Week 33, 2000 ,Svenson

Sjon



-- Previous week -- Most recent -- Next week --

-- Home -- Links -- To do -- Calendar -- The Gang -- The Undeniables --

Kelshon Saga. The logs. (book37.8 p221)

14-08 to 20-08

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxvii     Monday

 

2000-08-14

 


Nice dry and warm in the morning, well a bit too warm at 20°C. The day itself is less hot than yesterday but humidity is up. In the evening we get rain, all five drops of it.

I find no way to explain the extra records in the trigger file. So I log on to the Belgian system and start looking for localized versions of our trigger handling programs. But find none. So I go looking for other programs that might interact with the trigger file. And find nothing either.
So there I am, a day lost and not getting closer.


pd Redundancy : make sure you have multiple, if possible conflicting, definitions for every concept.

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxviii     Tuesday

 

2000-08-15

 


Still too hot at night with a few bursts of light rain, hardly enough to moisten the soil but the top temperatures are down to the middle twenties.

We have a holiday in Belgium (Ascension of Marry) so I stay home. That is one advantage of working out of the country. We have free on Belgian holidays because we work for a Belgian firm, but we also are free on Dutch holidays because we are employed at a Dutch company.

I have been busy almost the whole day clearing out some space and moving Caitlan to my main desk (still in the bedroom). And I installed Linux (Caldera eDesktop 2.4) on it.

That took the whole evening. I had the installation lock up three times. The main reason though was that I read a book (with Beethoven's 7th on the background) while installing so I paid not much attention when I had to enter things. I think everything works.
I am not sure because it is not on a network and I have no speakers connected all the rest seems to work.

I have got a KVM with serial mouse and AT keyboard but no sound pass trough. A cursory search in t'net didn't turn up any. I know KVM switches are intended to hook multiple servers up to one Keyboard-Video-Mouse and servers typically don't need sound. But there must bee enough people using these switches on multiple workstations to find some with sound support. Maybe I should wait and check Bobs hardware book


pd Redundancy : make sure you plan doubles of everything, like twice as much development time, and double budget, ...

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxix     Wednesday

 

2000-08-16

 


It's cooling down a bit (12°C) and most of the day looks like a typical Belgian summer : comfortably warm, lots of clouds with sunny breaks and a few light showers.

At work I have a fairly standard day as well. Or rather a quite day with no urgent calls and time to check some things out.

OMSI-3 is the old order entry system on the AS/400. Fully character terminal based so not at all 'modern'. It is efficient and quite performant as well but in this day and age it is definitively passé. Which is why TeleSales was developed. Originally TeleSales was a subset of the order entry program developed in Progress. From a fast and efficient subset it has grown to a full replacement for the order entry (losing its performance on the way). Of course today everything must be web based and e-named so an internet program was designed (well 'designed' (comment: censored) ... and became TOS2000. That is still being build.
So, now we are on the brink of having three order entry systems. And now it seems that all three use different business rules to determine which products can be ordered. As I said that is "design". On corporate scale.
Of course now, after the 'design' and when the implementation is all but finished the business rules must be determined.
And BTW OCE is also working on a SAP implementation so in a few years time we have yet another order entry system.

And I make a change to the transaction type file (well, to the maintenance program thereof). I should have done so before but I misinterpreted the problem report.

 

And I find a slightly disturbing inconsistency in the Open Linux installation. I did the install accepting all the defaults, except for the screen where I selected the LG 500LC from the list (being quite surprised to find my desktop LCD listed). I made two users, col and sjon. When I use a terminal windows (bash shell) I can only see the user directory for the user I logged on with. As it should, col cannot access the /root/sjon directory. When however I go via the "explorer" (Kfm ?) I can look into the directory for the other user.
Doesn't look good.


pd Redundancy : make sure you have a set of (empty) spare batteries

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxx     Thursday

 

2000-08-17

 


It rained most of the night, with some lightning early on, but past eight it gets dry. And apart from one more burst of rain it remains dry and even gets warm.
And now while typing this (at 22:55) I see the full moon rising, a bit hazy but in a black cloudless sky. Just a little mist obscuring the stars.

I pass the change to the Transaction file maintenance program over to the system test environment for approval.

Then I help Ronny at finding a problem with the orders that remained in the OH_P file. This is a temporal file used for passing unfinished orders (no real number yet) to and from the AS/400. In theory this file should be empty at the end of the day. In reality records remain in it if for some reason a program (either AS/400 or PC) gets canceled or if the network breaks. So some records are to be expected. In Belgium (why is it always Belgium?) there were 1700 records in it. They don't do any harm but they do indicate some lose ends, comparable to a memory leak.
After some prodding and probing (mental image of a dish of spaghetti and you have to compare all the strings to find a pattern) we notice that the transaction code is empty in all but one record. Somewhat later we find that, if the code is empty the header (OH_P) is filled for calculation but no calculation is done. The code can only be empty if the user actually clears it and the code is required, with a test just before saving. The only way records with empty codes can exist thus is if the user clears the code, requests a calculation and then doesn't save the order.
Later still I find two old problem reports; One that demands that order cannot be saved if the code is not filled. And the other that the code must not be filled for offers. The discussion between an offer and an order is only made at the time of saving.
And finally, after some consulting, we conclude that the code for is totally irrelevant for offers so whether it is filled or not makes no difference. So we must make the code mandatory at all time (with a good default).

Trough all that I didn't have time for reading all the Daynotes.

 

And I got into somewhat of a discussion with Tom (yes, The Tom or should I say Our Tom). Some time ago there was a discussion about search engines. With Tom indicating that he uses an external service because (among other reasons) he doesn't want to run CGI scripts on Hydra. Which I think is a valid reason.
He is also running the Seti@home program on his computers, including Hydra.

When I look at that situation, from the programmers side, I see an inconsistency. Some unknown programs (CGI programs) are not allowed to while other, equally unknown programs are allowed to play. Looking at it (as a programmer) both Seti and CGI programs are just that, programs (Seti is coded in C but distributed as binary, CGI programs can be almost any language). A well written CGI program is just as safe as any other program running on your box on top of that, most (but definitively not all) CGI programs, and certainly the scripts, provide their source so they can be checked. Ok, that is not trivial. But then Seti doesn't disclose it source (they say why on their site) so cannot be checked
When he looks at the situation he sees things from an administrator/opperator side (and we all know that programmers and sys-ops have very different and opposing views (except when looking at users of course &l;g>). He knows the Seti program in the sense that he has ran it for years without ill effects or misbehavior. And the Seti organization is a well organized and disciplined one, not set at maximizing some revenue. In a word he trust them and their program. He doesn't know the ins and outs of CGI programming and he knows (rumors and facts) that CGI scripts can pose security treats. So he doesn't trust them.

This clearly is just a difference in view (programmer versus sys-op) and not a difference in philosophy (Tom versus Sjon).
The proof that, when I don my sys-op hat, I apply the same philosophy as Tom?
I don't run a search engine even though I do have CGI ability on the hosting server. So I could copy a program and script and run it. I don't know how these things work so I don't do that. Yet.

 

Aha. It is not often one sees the AS/400 mentioned in the press, so this Register article jumped out to me. And it sounds like good news as well. Why, I do make a living working on AS/400 programs.


"You are cured" says the psychiatrist, "you don't have to come back anymore."
"That is a pity" replies the client, "yesterday I was Napoleon and now I am nothing anymore."

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxxi     Friday

 

2000-08-18

 


Dry and warm (14°C) in the morning with just enough and the right kind of clouds to show off in long red bands while the sun laboriously climbs over the horizon. And it stays that way until after 20:00 when a slow rain descends.

Theo had a discussion planned in the morning about passing over the TOS2000 system from the Progress people to Ronny and me. This should be done somewhere in September.
Sasha, one of the developers from Progress doing the TOS2000 programming (lead developer actually) had a small problem so we had to go and help him on a bit. Ronny has been doing that all week but this time he needed me because the problem seemed to lay on the AS/400 side. Well it wasn't but logging the communication coming on the AS/400 we found the problem. Something with a before and an after transaction trigger. A quick and dirty solution was possible but Q'n D solutions have a habit to bite back later and later could be when the current developers were gone so that we would have to solve it we pushed for a good definitive solution.
And we got that, at the cost of extra work for Sasha. <eG>

Afterwards (afternoon) we had the planned discussion. Which did not lead to a conclusion but it did cause Theo to notice some upcoming problems that were not (yet) on his horizon.

And then, to stay in style, the Belgians pulled another stint.
During week 25 to 29 I worked on a correction program for the problem with archived orders in status VAL. A very tricky program to test. I also adapted the actual archiving program to prevent the problem from recurring.
The danger was that the new archiving program, when run before the correction program would irrevocably destroy the records needing correction. Upon delivery I added a description of how the programs should be used, explicitly stating (twice) that the correction program had to be run before installing and using the corrected archiving program. Later the operator, Robert, from Belgium called and I walked him trough the process on a copied set of data. All went well.
Now guess what they did? On they active production database.
Ha, I see, nobody needs a second guess.
So now they lost their data and there is no way to get it back without destroying several days worth of work.

It's like giving someone a hand grenade with the instruction, in 24 point Arial, to 'pull out the pin and throw the grenade'. And then seeing them throwing the pin.


Since the accident with the Concord recently,
the Russians have decided to keep all their submarines grounded.

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxxii     Saturday

 

2000-08-19

 


It started to rain yesterday evening and it continued to do so till early afternoon. At times it was only drizzling. Now its cool and dry.

Normally, on Saturday I sleep in till about 8:00. Because Suzan isn't home however I had to get up at 6:00 so I had a few extra time for cleaning up the kitchen.

Shopping. The baby-AT main board has arrived so I pick that up along with a 64MB DIMM and a floppy drive. I should have picked up a CD drive as well but forgot.
The rest of my shopping only netted food and clothes.

Shopping was a bit short skirted because I had to prepare lunch and pickup my father again from hospital.
Lunch was steak with haricots, fresh from the garden, and young potatoes, dug up yesterday. Decorated with a good pepper sauce and washed down with a good Australian red wine. And finished off by a ... long siesta.
BTW on Saturday and Sunday Lunch is the main meal here.


Bugs are social creatures, you will never find them alone.

 

Top <<<     Mon -- Tue -- Wed -- Thu -- Fri -- Sat -- Sun     >>> Bottom

MM-ccxxxiii     Sunday

 

2000-08-20

 


After a rainy night we get a warm and sunny morning, with clouds drifting in in the afternoon. Followed by rain, lightning and far off rolling thunder.

With 16°C at 6:00 it was a bit too warm for running. Some weeks ago I took another route which turned out too short. This week I took yet another route, this time at least as long as the usual one. But with two bridges crossing the railroad it's more taxing.

I tried to install the new main board in the old AT case and while it fits the old plastic supports didn't fit the right holes so I had to move them around. The box is quite old and the supports turned out to be very brittle. So now I don't have enough supports left. I also realized that this case has only one 3.5 bay so I need a set of rails to fit the hard disk in one of the three 5.25 bays.
Bad planning.

And later I hit a Linux wall.
I have Caldera's eDesktop installed on Caitlan using all the defaults. And now I cannot get at the CD drive nor at the floppy drive. When I try to mount them the drive light comes on for a second and I get a kfm window saying "mounting /dev/fd0 ..." which stays up indefinite. When I try to mount via the KDE Kontrol Center, file systems dialog I get "No kernel driver for system auto available". And of course the online help doesn't help (stole that feature from Microsoft) and the manual isn't clear either.
Sigh.


pd patience : the mind set you don't have when you need it.

 

 
-- Home -- Links -- To do -- Calendar -- The Gang -- Previous week -- Next week --

Swijsen © 2000

A day you don't learn something new is a wasted day.