Re: 'page impression' techniques for banner ads

This WebDNA talk-list message is from

1999


It keeps the original formatting.
numero = 22217
interpreted = N
texte = I once (with WebCat 2.x) tried doing replaces inside a HUGE database file (>100 Meg RAM). Even with autocommit off, it occasionally flushed the whole .db to disk. This made approach #1 unreasonable. Perhaps it is OK with much smaller .db's where the occasional flush to disk is not noticeable. At the time, PCS confirmed that replaces do occasionally get flushed to disk even with autocommit off. I believe they said around every 100 operations or so.Has this changed?I went with appending a separate database instead. But I'm curious how the speed of appending a database varies with size of the .db. If I let this log file grow, will there be a speed hit as it continues to append the file?ThanksSandy>Yes, I think your ordering is correct, and you may even be able to speed up >the replace a little with case-sensitivity turned on. Replace operates >entirely RAM-based as often as possible. > >>#1 would be fastest because it uses a database to calculate and store the >>total page impressions. If replacing a database field value is always >>faster than writing that value to a separate text file, this would seem to >>be the fastest solution -- because it does not require a disk hit. >> >>#2 would be second-fastest because all it does is append a single >>character to the end of a separate text file -- no additional [include]s, >>and no [math] to perform ... >> >>#3 would be slowest because it must include the value of a separate text >>file, then perform a math calculations on that value, then re-write the >>results to the separate text file. > >Technical Support | ==== eCommerce and Beyond ==== >Pacific Coast Software | WebCatalog, WebMerchant, >11770 Bernardo Plaza Court | SiteEdit Pro, PhotoMaster, >San Diego, CA 92128 | Typhoon >619/675-1106 Fax: 619/675-0372 | http://www.smithmicro.com/ > > Associated Messages, from the most recent to the oldest:

    
  1. Re: 'page impression' techniques for banner ads (PCS Technical Support 1999)
  2. Re: 'page impression' techniques for banner ads (Sandra L. Pitner 1999)
  3. Re: 'page impression' techniques for banner ads (PCS Technical Support 1999)
  4. Re: 'page impression' techniques for banner ads (PCS Technical Support 1999)
  5. Re: 'page impression' techniques for banner ads (Kenneth Grome 1999)
  6. Re: 'page impression' techniques for banner ads (Brian B. Burton 1999)
  7. Re: 'page impression' techniques for banner ads (Brian B. Burton 1999)
  8. Re: 'page impression' techniques for banner ads (Kenneth Grome 1999)
  9. Re: 'page impression' techniques for banner ads (Kenneth Grome 1999)
  10. Re: 'page impression' techniques for banner ads (Kenneth Grome 1999)
  11. Re: 'page impression' techniques for banner ads (Brian B. Burton 1999)
  12. Re: 'page impression' techniques for banner ads (The Mooseman 1999)
  13. Re: 'page impression' techniques for banner ads (olin 1999)
  14. Re: 'page impression' techniques for banner ads (PCS Technical Support 1999)
  15. 'page impression' techniques for banner ads (Kenneth Grome 1999)
I once (with WebCat 2.x) tried doing replaces inside a HUGE database file (>100 Meg RAM). Even with autocommit off, it occasionally flushed the whole .db to disk. This made approach #1 unreasonable. Perhaps it is OK with much smaller .db's where the occasional flush to disk is not noticeable. At the time, PCS confirmed that replaces do occasionally get flushed to disk even with autocommit off. I believe they said around every 100 operations or so.Has this changed?I went with appending a separate database instead. But I'm curious how the speed of appending a database varies with size of the .db. If I let this log file grow, will there be a speed hit as it continues to append the file?ThanksSandy>Yes, I think your ordering is correct, and you may even be able to speed up >the replace a little with case-sensitivity turned on. Replace operates >entirely RAM-based as often as possible. > >>#1 would be fastest because it uses a database to calculate and store the >>total page impressions. If replacing a database field value is always >>faster than writing that value to a separate text file, this would seem to >>be the fastest solution -- because it does not require a disk hit. >> >>#2 would be second-fastest because all it does is append a single >>character to the end of a separate text file -- no additional [include]s, >>and no [math] to perform ... >> >>#3 would be slowest because it must include the value of a separate text >>file, then perform a math calculations on that value, then re-write the >>results to the separate text file. > >Technical Support | ==== eCommerce and Beyond ==== >Pacific Coast Software | WebCatalog, WebMerchant, >11770 Bernardo Plaza Court | SiteEdit Pro, PhotoMaster, >San Diego, CA 92128 | Typhoon >619/675-1106 Fax: 619/675-0372 | http://www.smithmicro.com/ > > Sandra L. Pitner

DOWNLOAD WEBDNA NOW!

Top Articles:

Talk List

The WebDNA community talk-list is the best place to get some help: several hundred extremely proficient programmers with an excellent knowledge of WebDNA and an excellent spirit will deliver all the tips and tricks you can imagine...

Related Readings:

Linux uninstall / reinstall (2003) [SearchString] problem with [search] context (1997) Shipping costs again (1998) Plugin or CGI or both (1997) Rendering out a page-test results (1997) Cart # Starting at 1000 (2000) WebMerchant -- Sweeping WebDelivery < 24 hrs (2000) Shopping Cart - prices? (1997) Mult-File Upload w/ Web Star (2000) Web Catalog Instability (2000) OK, here goes... (1997) emailer (1997) E-mail Attachments (1997) Help! (2000) WebCat2 - [include] tags (1997) Server IP address? (1998) syntax question, not in online refernce (1997) upgrade? (1997) syntax question, not in online refernce (1997) Out of the woodwork (2007)