Search results

you're drunk.

The Progress conference is called ProgressNEXT this year. This site and communities.progress.com are good places to ask questions. OEHive.org is a good source for code samples and tools. Progress has a program of on-line education, which I have no experience with.

You need some interaction with the recipient ... what do they actually want you to do with the new lines? If they want to preserve them, then mutually pick a character to substitute and replace on export and reverse replace on import. If they can't actually handle them at all, then decide...

FWIW, if the data can be safely held in memory and you don't need an index, i.e., order doesn't matter, work-tables are higher performance than temp-tables and consume less space. It is possible to build a work-table in order, but if you need order just use a temp-table.

You need to look carefully at the various ways of passing temp-tables. Passing as a copy means clean separation, but it also means the overhead of making a copy. Passing by reference just passes a handle and so is much more performant, but means you do not have the separation. Whether or not...

Note, the two are independent. DECIMALS determine the precision with which the number is stored. You have no control over the 0s and 1s in that storage, but it will store enough bits to give you the desired decimal precision in what is stored. You may then display that number with any...

A shared temp-table is almost certainly the wrong solution. Much better is to generate it once and store it somewhere - in a temporary table, possibly in its own DB freshly created by the process which does the generation, or as XML or JSON. Then, have the consuming programs get it from that...

And now you have a tool for the next time. For formatting code, use the icon between media and drafts, select code, and then paste your code in the inside. Helps a lot.

Well, you seem to managed to fork the post anyway ... You might try being a wee bit more descriptive about what your problem is so that we have at least the slightest clue.

It would help a lot if you put code tags around that so that the indenting was maintained. A slow query is almost always a question of bad indexing. The file names look like MFG/Pro, but I don't have the dictionary for that. A COMPILE XREF would show you what actual indexes are used and you...

Why would you expect these to be the same. In the first on you used two fields to make the selection; in the second you used one. In the first you are ganging together three FIRST-OF, which is obviously not doing what you think. On the very first record, all three fields will be first-of, but...

There is no table or log that will give you an accurate user count. For starters, there are 4 different types of user -- named, registered client, concurrent, and access agent. Access Agent can only be used for users which are not individually identified and use the system less than 2 hours...

And, compile with the listing option to see where the transaction blocks are.

First, you need to define "user". Is it going to match PSC's definition of Named User or Registered Client or something else. Then you need to implement logic and database that reflects that definition and provide management tools including how to notice that a user has disconnected in some...

Have you considered other mechanisms altogether? Like populating a temp-table and then exporting it as JSON? So, are you now saying that there is DB access in this loop? And suggesting that the concatenation is slower than the DB access?

I gather from the ~n in the commands that this is intended to be a multi-line output? Is each line different?

Well, I'm pretty confused about what you are doing ... why is there a 10K iteration which contains only some form of concatenation? Do you really start the loop with 10K values available?

Something doesn't make sense. I can't see a single substitute command taking 3-5 seconds to complete.

Yes, but again, what else is happening in each iteration of the loop. If the time consumed in the other parts is substantial, as it will be if you are doing anything with a file or the database, then the speed difference of these two options is meaningless. And, if you have UI between each...

My point is that if the real loop that you run 10,000 times is actually something like this: compute file name open file put something in file close file then the compute file name time will be a tiny fraction of the total time spent in each pass, small enough that you won't even reliably be...