Network Operation Performance

[Jim Coltz]

Re: Network Operation Performance

I would like to just put in a plug here for Ira's CSDA Code utility. I am using this now to help speed up my app and have made some major speed improvements. One in particular is in regards to the a5_get_records_in_query() function as a short cut to opening a table or set and running the query conventionally. What I have found is that going the conventional way, although much longer to type out than the function, results in a significant improvement in speed. I don't remember the actual numbers as I am writing this from where I don't have access to A5, but I recall it definitely being a "wow" factor. I have since started to pick through the rest of the code and optimize where I can. Little by little this app is starting to fly where it once only ran.

[Bob Whitaker]

Re: Network Operation Performance

Could I just add my thanks to Ira for taking the time to point use mere mortals in the Alpha firmament in the right direction.

Bob

[csda1]

Re: Network Operation Performance

Jim:
Thank you for your kind comments on the CSDA_CodeUtility. It was designed for my use to speed up MY Development and to speed up MY execution times. Then I figured some of my time and expense of it's development could be recouped by offering my tool to the A5 community.

I believe it will save any A5 developer enough time to recoup the expense of the product in 1 or 2 man-days of Alpha 5 work, which is a great payback for any tool! Of course, each person's mileage (or kilometerage) may vary.:) )

Jeff:

For what it is worth, Foxpro 2.6 works very similar to Alpha - but it had none of the great features like sets, data dictionary etc. Although, I always thought it only tried to lock records while they were being edited. Maybe I was wrong about that after reading your post since they are the same underlying data engine.

If they use identical database structures, there is a limit to the speed based upon the lock/unlock issues. I believe A5 and Foxpro have pretty much hit those limits. Code execution is another issue. XBasic is interpreted of pseudo-compiled code. It is not as fast, but if you use the right functions can be. Alpha 5 has most of the key functions needed coded so they go as fast as compiled code. It's only the higher level Xbasic or Action Scripts that typically slow it down.

But, you could go against the tables with SQL statements and return the data you want for queries blazingly fast. One nice thing Foxpro did do was that you could run a query with SQL statements it created a cursor with the result. The cursor was basically a temporary table built from your select statement but contained in memory. You could browse this cursor, run reports on it etc. I miss that.

A5 can do the same. In fact, the MySQL interface is one of (if not the #1) fastest connection.

It seems Alpha's queries are really just filters? I have paid great attention to making use of the LQO capabilities of Alpha. Foxpro's word for LQO was Rushmore Technology.

Mount Rushmore & LQO are the same. Queries are filters applied after LQO process is performed.

It would be nice to see a compiler from Alpha. Do you know if there has ever been any talks about doing so? Foxpro had it and it made updates a breeze.

I Know Alpha has experimented with code compilation, which would speed up XBasic code's execution by a very large factor (but would not affect the speed of a query or other A5 functions that are already highly optimized). There are issues with many special functions like eval(), *for_each() etc in designing a compilation. Nevertheless, many of these could be overcome, although probably with great pain.

Compiling everything into a true exe like Clipper is probably not ever going to happen, because it does not gain you much (except 1 file vs a bunch of files to install) but I could be wrong.

The Alpha 5 code method is basically taking complex database operations, eg. crosstab, summary, query etc) and providing a high level (XBasic/Action Scripting) interface. Probably less than 5% of the time is spent in the running of the actual XBasic code (well, if written correctly anyway), and that is what a compiler would improve, not the lower-level underlying function code that Alpha has provided.

[jkletrovets]

Re: Network Operation Performance

Ira

A5 can do the same. In fact, the MySQL interface is one of (if not the #1) fastest connection

I saw where you could do this in the desktop Ver 8 but I didn't think you could do it in Ver7. This will be REAL nice when Alpha gets the native desktop forms to work with the client/server databases! I know you can do it now in ver 8 with Xdialog but I haven't played with that to see if it is a pain or not.

XBasic is interpreted of pseudo-compiled code. It is not as fast, but if you use the right functions can be. Alpha 5 has most of the key functions needed coded so they go as fast as compiled code. It's only the higher level Xbasic or Action Scripts that typically slow it down.

When you refer to "key functions" are you referring to those that use an index in the function like using dbcount() over tablecount()?

Thanks again for taking the time to share your experience and knowledge. Your thorough and in depth explanations are very helpful.

Regards,

Jeff

[csda1]

Re: Network Operation Performance

Hi Jeff,

When you refer to "key functions" are you referring to those that use an index in the function like using dbcount() over tablecount()?

When I talk about key functions (as opposed to functions that use keys - in other words index keys), I mean the function that have been coded in really efficient C code that do a function. How can you tell? Measure the speed of the function, and you will see a wide difference between XBasic coded functions and methods versus C coded function and methods.

If it seems like an XBasic function sort of simplifies use of a more simplistic function, it is probably an XBasic "Wrapper", or in other words, a function written in XBasic that does some stuff before calling a lower level function.

E.g LINE_COUNT(x) probably is written as an XBasic function whose code is W_Count(x,CRLF()). The call overhead doubles the time.

[jkletrovets]

Re: Network Operation Performance

Thank you once again Ira. Looks like I need to get your utility for the timing functionality as well as all of the other features. Will do so this weekend.

Best regards,

Jeff