|
Previous calun < Tgu | Sal | Gli | Ubo | Mir | Ble | Suf | Lie | Utr > Next calun |
|
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
Freezing early in the night indeed. By morning however clouds and rain and high temperatures have arrived turning the evening promise of a good day into a muddy grey, miserably drizzling affair. Enough dry spells to break the monotony though. Together with Koen I keep looking for the problem in the copy-from-consumption. I pruned down the data set, throwing out thousands of records, and modifying others to make sure there is at least something to show (and something that shouldn't be shown). The problem is that we both don't see the problem and every test ends in an endless loop. If we let the loop run wild too long it eats all the performance from the NT server making it difficult to even kill the process like what happened yesterday. Not all courts are impressed by strong-arm tactics. Maybe others will now look different at BSA requests. www.theregister.co.uk/content/4/24115.html I had booked half a day off to pick up my car. Still, I got to the garage in time (seconds) to pick up the car. I then went shopping (needed bread and dinner) and got home about an hour earlier than usual. (Yes, leaving work three hours early results in one hour extra home-time. Sigh.) Tuesday 2002-02-18
208.055 Tguridare
^
Current temp is 11.0 C in and 6.7 C out.
In the end, after getting rid of a few other, lesser, problems we pinpoint the problem. We have set a loop to end when no fitting record is found. When the end of the file is reached Progress does not report that as 'no record found' but as something else. So the loop doesn't stop.
That may sound an easy and common problem, and it is. Only Progress is a 4th generation language so every action is wrapped in some layers and placed in black boxes. And when you are not familiar with the black boxes and wrappers the obvious can not be so.
Well it needed its regular 40.000 km service and I needed a new wind screen cause the old one was cracked after making intimate contact with a stone. My fathers car is still here so I drove that to work for two days but I had to be back in time, before 17h00, to pick up the car. I must count on a two hour commute so I had to leave work at 15h00 at the latest, taking half a day of make sense then.
Actually half a day off doesn't make sense because then I am commuting longer than working, I should and would have taken a whole day off but there is a bit much work at the moment.
So I booked half a day off and then forgot to go home at noon. I only remembered at about half past three. Oeps.
So dinner, pork-rib with Brussels sprouts was taken at leisure.
Time is never working for you, it is always working against you.
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.056 Saldare
^
Heavy rain and strong wind rule the day. With some hail and some sun being the interesting interludes. Basically the same as yesterday. I like it though.
Current temp is 11.6 C in and 1.7 c out.We finally worked out where the problem was in the copy-from-consumption program. Apparently the 'FetchNext' routine will pick up the next record from a query set. If it reaches the last record of the set it will return the data and an indicator that that is the last record. If the routine is called again it will not throw back an error but just return the last record again (and reporting again that it is the last record.
We are used, from RPG on the AS/400, that trying to get the next record after the last will result in an error (which is trapped). Of course expecting Progress to react as logically as RPG is stupid and yet we did.
After correcting that I noticed a problem in the passing back of the selected items to the order-entry program. This was caused by the same phenomena. However it occurred in a routine and using a file that hasn't been changed for almost a year. It is easy enough to solve now that I know what to look for.
So now the copy-from-consumption is rebuilt for enhanced performance. And it's working correctly.One of the reasons why it took so long is that we cannot run a debugger. All we can do is add messages, compile and run the stuff and then check the agent log. If two people are working and testing things in TOS the agent log becomes mixed up. Especially problematic if you realise that the tool to view the log locks that log. The logging goes on but if the tool runs in two sessions only one session will update automatically while the other will only update sporadically. So in the early stages I didn't do much 'debugging' because Koen was doing the performance measuring via the log.
Later, Koen started to look into the problem as well. Quite helpful but obtrusive as well because it is not a good practice to have two people making changes to the same source at the same time.Then there was a quicky. In Austria some problem occurred in the background queue. The problem was not my affair, Frank helped them out with it. The result however was a missing Trigger record, and Frank didn't know how to fill it manually. So I did (after checking that it was indeed missing and needed).
And Theo came up with a request from Germany.
Sometimes TeleSales will crash because the connection to the NT server gets interrupted. That is a known problem (but there is no known solution) in the Progress environment and has nothing to do with our own programming.
The result of such a crash is that the user loses the work he was doing on an order. Well, the work isn't actually lost, the temporal order remains available in the Work_Oh and _OL files but there is no way the user can retrieve it. (There is always a way of course but normal users can't and shouldn't go that way.) So now Theo wants a new module so that the user can retrieve the lost data.
My proposal was to provide a program where the lost data is converted into a normal (held) order that the user can pick up with the standard existing tools (select-order and order-entry). That is the easiest and least intrusive way.
Koen proposes a method to 'resume work' directly via order-entry without converting the temporal order into a real order. That prevents clogging up of the order files.
I guess he will get his way (he always does) but I am not sure he will get it worked out. The problem is that his solution means additions must be made to order-entry and that program has been hitting the compiler size limits. Ronny had to remove all comments and use a bunch of tricks to fit it in. Adding extra coding, even if it is simple coding, could be a problem.
(And preventing the real order files from being 'clogged up' is, I think, a stupid concept. I suppose that if the user picks up the lost data it is his intention to actually save the order so the data ends up in the order files anyway. )
Home extra late, because I ... forgot to keep an eye on the clock in the office. So I made a quick diner of some mixed salad (seasoned with some basil, olive oil and sherry-vinegar) and cold fish.
Time is always working against you, but if you are lucky it is working harder against your opponent.
Wednesday 2002-02-20
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.057 Glindare
^
It rained a lot during the night and in the morning. There must have been a clear period however because my car was covered in frozen raindrops. Apart from a few bouts of rain it's a dry and even sunny day. So, Dan, I actually know what sun is.
Current temp is 11.1 C in and 2.7 C out. But the way the temperature is jojo-ing these are not reliable numbers.I did pick up the translated messages for Italy again. And got them loaded without incident. Well, there was actually a problem of course. The Italians use a lot of apostrophes and that causes problems if the text is used to dynamically compose HTML pages and JavaScripts, an apostrophe (or quote) in the wrong place wreaks havoc with variable delimitation. I had to replace all their inline quotes with % . (Which reminds me I have to do a similar exercise on the AS/400 database because the item descriptions and addresses and names contain quotes, double quotes and other disruptive stuff.)
The merging of the translated messages with the other messages was not really difficult but I hate to follow instructions blindly. I like to know what I am doing. Which is why I write down ('type in' actually) things into a sort of guide/manual.And that is where I did spend some more time on after the translations. The documentation is growing steadily, I am over 200 html pages not including the navigation menus. Because this is all lovingly (ahum) hand crafted HTML and because it is made without prior planning the structure is not conductive for finding things back. Well, I find my way in it because I wrote the lot but I don't think anyone else would have an easy job finding the right stuff in there.
Well, that was yesterday, today I made a kind of map (not finished yet), laying out all the links so finding things now has become a piece of cake (OK a cake with lots of nuts baked in and no icing yet). Of course the documentation keeps growing as I add new stuff so I guess it will become un-organised again in no time.
After the standard shopping run I made a bami dish (noodles with vegetable and spices). Not the best I have ever made but good enough
When I check my keyboard and mouse I wonder why they aren't sold as cleaning tools.
They attract dust and dirt so that the rest of the desk remains shiningly clean.
Thursday 2002-02-21
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.058 Ubodare
^
Loads of wind and lots of rain (and some hail and a few flashes of lightning and ...). The small drop sized type of rain that, combined with wind gets into places where no rain
has gone beforeis supposed to go. While it may sound ugly I actually like this type of weather. Of course this is combined with a completely grey covered sky.
Current temp is 12.4 C in and 4.0 C out.It turned out that the prolem we have been having with the copy-from-consumption page occurs on most of the pages. After searching and experimenting (and a call to Sasha, not verry helpfull but more confirming our findings) we find that there is a problem in Progress.
We are using SDOs (Smart Data Objects) to access the database. When you initialise an SDO, which you do when a program starts, you indicate how many lines (records) from the database must be stored (cache like) in the SDO. When you use the file randomly, for example to retreive single descriptions you set that to 1 because you will requet only one record at a time. When you fill up a list you set it higher, for example at 25 when you intend to show the records in groups of 25. You retrieve the records one by one from the SDO (fetchFirst, FetchNext, ... routines). The SDO will return the requested record and an indication showing what record you got (Firstrecord, LastRecord, NoRecordFound, NotFirstNorLast, ...).
To increse the perfromance all the SDOs gets loaded and started at when the server process (agent) is started and they remain active till the process stops. That means that all the programs use the same SDO. Before that change each program started the SDOs that it needed and delete the SDOs again when it finished. They were private and now they are common. The result now is that for example program-1 initialised an SDO for 20 lines. Then another program initialises the same SDO for 1 line. Then the first program (which never stopped) requests a record asuming the SDO is still set for 20 records (remember this).
When an SDO is set for 20 record and the query only nets 4 records on the first request the SDO will retreive all 4 records at once. You read them consecutively from the SDO so the returns are 'FirstRecord', 'NotFirstNorLast', 'NotFirstNorLast', 'LastRecord'. So when you receive the 'Last Record' indication you know that the returned record must be processed and that you don't have to request another one. When you do the same when the SDO is set to one record the SDO will fetch only one record from the database. The first record is returned as 'FirstRecord'. On your next request the SDO has to go back to the database finds one and returns it. You get 'NotFirstNorLast'. On your third request the SDO has to go back to the database again finds one and returns it. You get 'NotFirstNorLast'. On your fourth request the SDO goes back to the database finds one and returns it. Because the SDO doesn't look further it doesn't know this is the last one, you get 'NotFirstNorLast'. Of course you do yet another request, the SDO goes to the database and oly now notices that there are no more records. Here the bug hits, in stead of saying there are no records the SDO returns the previous record (number-4) and indicates this is the 'LastRecord'.
So you receive the last record once or twice depending on the setting of the number of SDO lines. And you remember that that setting is no longer fixed, it can change while the program is running.
Bother.
Big Bugging Bother. (B3)
Shows that police do actuall have a sense of humour www.arcataeye.com/police/020219police01.shtml
(link found over at Dan's TimeSink)Trying to get the internet bottled up again? www.silicon.com/a51509
The European take (proposal) on software patents looks a lot more reasonable than the American 'everything is pantentable' view. Of course there is no telling how the rules will get interpreted. www.silicon.com/a51510
Nothing surprising but the title garbbed my attention: "Mobile radiation shields as much use as chocolate teapots" (does anyone like IceTea?) www.silicon.com/a51483
I forgot to come home in time and only left the office after 19h00. So I am home hoplessly late. Too late for dinner anyway.
Yeah, too late for dinner but not to late to be extremely wordy for a second time this week. Sorry about that but if you don't like it you can always check the length before starting to read. That is why I have the word-counter up under the date.
Some people address 'bugs' as feature.
A bit like the polsh on some torture tools.
Friday 2002-02-22
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.059 Mirbandare
^
Clear and dry in the morning and during most of the day. Sadly enough 'most' is not 'all'. There were frequent patched of rain keeping the garden unworkable.
Current emp is 14.3 c in and 3.7 C out.Shopping. Normal stuff like wine and fruit and vegetables.
In the afternoon I help out my brother with some chores that need an extra hand like lifting out (and putting back a front door).
I also help out while he is taking the gearbox from his bike apart. That bike is actually my faters old bike. And I mean 'old', it was bought shortly after the war (WWII that is). It has 3 speeds with the gears inside the rear wheel axle, not such a newfangled external thing. It also got drum barkes fron and back.
A small rod inside the gears was broken. The bike shop did have a replacement for the rod but some other parts can't be had anymore and the bike repair man didn't dare to open up the gearbox for fear of breaking one of the unique rings and plugs. We succeed in opening and dis-assembling the gears without breaking anything.
slurg
Saturday 2002-02-23
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.060 Blegidare
^
Very unsettled weather, going from cheery and sunny to rain and stormy (and back) in a metter of minutes.
Current temp is 13.7 c in and 2.0 C out.
Running. I start of in dry weather but before I leave the street it a srong wind picks up and it starts to rain. Well actually its dense drizzle. And it is frozen. Each droplet stinging like a needle. Later the frozen rain shifts into snow and back to rain. And when I get back home about an hour later it has stopped raining.
Most of the day remains unsettled and wet so I stay in and keep myself busy. Cleaning, cooking, lazying, etc. Just the normal weekend resting stuff.
Peter got the bike's gears reassembled and working again. But he is not going to test-ride in the cyreent weather.
plitch, plitch
Sunday 2002-02-24
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.061 Sufirdare
^
It rains. Morning, noon and evening, with a strong gusty wind blowing as well. While it rains all day that rain is not constant, it varies between downpour and drizzle. Quite interesting.
Curremnt temp is 13.5 C in and 8.4 C out.The TOS problem we pinned down last Friday cannot be solved by us. It is a bug in the Progress programs, in the query.p super-procedure. Koen tried to change the program, putting messages for the log in for good measure. The first message, in the initialisation routine is reached but the actual routines (FetchNext and FetchFirst) also have unconditioned messages but they don't show up in the log. As if the program loads but then gets replaced by an other, older copy. Koen did call someone at Progress but he didn't give an answer, someone else will call back tomorrow.
If we don't get an answer by then we will have to hunt down every instance of query.p on the network and disable/replace them all. There is definitively something weird and inconsistant happening.
"These ventures look bad, smell bad and sound bad," Napster judge says www.theregister.co.uk/content/6/24178.html .
Mabe RIAA is now the new official spelling for greed.Well, well, well. MS isn't having a roll over in the PDA market. www.silicon.com/a51537
"Microsoft does best when behind, once they catch up they don't break ahead". Appatrently MS thinks it has caught up with Nokia and co. www.theregister.co.uk/content/54/24145.html
IP6 is needed. And maybe sooner than expected. www.silicon.com/a51525 (and the finger-pointing begins).
Home late, too late for a regular dinner anyway.
If the attention span is too short, you just repeat the message.
(Typical Inquiry humor www.theinquirer.net/22020216.htm )
Monday 2002-02-25
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.062 Lielidare
^
It rains. Morning, noon and evening, with a strong gusty wind blowing as well. While it rains all day that rain is not constant, it varies between downpour and drizzle. Quite interesting. And if you think this was just a copy of Sufir days weather, it is not. There was about 10 minutes sunshine. Falling on my screen so I had to close the blinds.
Current temp is 14.2 C in and 6.6 C out.After some calls to progress (yesterday and today) Wilbert prepares the Sports database for us to try to reproduce the problem there. The only way to convince the people from Progress that a problem is caused by a bug in their programs is to demonstrate it on the Sports database. If you don't show that it goes wrong on that they keep blaming your own programs and/or database.
The Sports database is part of the Progress installation intended to showcase the Progress capabilities and ease of development.
After preparing the Sports database first Koen and then I have a go at it. Neither of us succeeds in making a working program from scratch, never mind reproducing the problem. Well, I do get a Progress "program" running but we need a webspeed program and I don't get the generator to find/recognise the <!--WSStart ... tag. Without the generated triggers I can't get a table filled. I notice that some of the triggers are generated but they don't show up in the outline view of the program and can only be found in the actual source. I haven't got a clue about what is wrong, most of the actions are done by wizards.
Yeah, and we all know how temperamental wizards can be.
Shooting oneself in the foot. With a nuke. www.theregister.co.uk/content/4/24192.html You can count how long StarOffice survives after Sun starts charging for it.
I don't think you will need two hands for counting.
And you can count on MS rejoycing and pointing out the 'failure' of everything about open source.Quick dinner, Spagheti with a spiced up Bolognese sauce.
Waking up problems in the morning?
Maybe you should check and reset your BIOS.
(as in Biological Information Overload Safety)
Tuesday 2002-02-26
Prev <
055 |
056 |
057 |
058 |
059 |
060 |
061 |
062 |
063 >
Next
208.063 Utrodare
^
It still rains. Morning, noon and evening, with the same strong gusty wind blowing. The rain is less heavy and the sun is more active so there is a change. Now we have a full moon and falling (towards a more normal level) temperatures.
Current temp is 14.2 C in and 4.0 C out. (about half an hour ago that was still 5.7 C out)I finally got Progress to behave as I expected. The problem was that the wizard was ignoring the <!--WSStart ... tag. This is standard behaviour, the WSStart and WSEnd are one of the custom tags, I knew we had custom tags but I didn't know which ones. I found the list(s) with the tag definitions, located all over the place actually, and I copied one of them to the directory of the sports database. Which (of course) didn't work. I don't know where I must indicate to the wizard which tag-definition file he must use and that information isn't in the manual (well, I haven't found it yet).
I ended up changing the tag definition in the original file (on my local disk) and now it works. Of course that isn't the right place to put the change but I'll find out later how to do it correctly. Now I don't have the time. I just hit the next problem.
Which is testing the now finished test application on the sports database. Of course there is no predefined entry page to point a browser at.What I find the most frustrating thing with this Progress development is that it goes slow. More than half the time (when I know what I am doing) I am waiting on the system. The hourglass has no secrets for me anymore. All that waiting is taking the fun (and motivation) out of the job.
Why isn't the maker of the table in court on accounts of manslaughter. It was after all the table that caused the fatal injury and not the Nintendo box. www.theregister.co.uk/content/54/24220.html
Again some companies that are discovering the joys of 'partnering' with Microsoft. www.theregister.co.uk/content/4/24221.html
Untill now CRM did not stand for Creative Revenue Maximisation.
Home regular (which is late of course). I had planned on a steak with chicory but once home I didn't have much of an appetite for steak so I just made some rice mix without meat.
for comments -->
There used to be only one rule in war: When you win, you get to make up the Rules.
Now, whether you win or lose, the media are making the rules.
Wednesday 2002-02-27
|
Previous calun < Tgu | Sal | Gli | Ubo | Mir | Ble | Suf | Lie | Utr > Next calun |
|