Re: Performance issues
Karr, David wrote: Just in case, are you doing your measurements AFTER the first generation/compile of the servlet? It's not meaningful to measure taglib performance before then. Measurements are taking place during a continuous load of approx 60 users. With respect to caching evaluation results, it's probably not worth it, as it would have to do most of the evaluation before it could determine that nothing has changed, causing a cache read. OK, thx for the reply Colin -Original Message- From: Colin Chalmers [mailto:[EMAIL PROTECTED] Hi all, We've just started implementing TagLibs in our App and are seeing a couple of performance issues which I'd like to check with the group. We make heavy use of theExpressionEvaluationManger, is there any way to tweak this so that it pools/caches results? We also make a lot of use of the choose/when/otherwise Can this be a potential bottleneck compared to using a simple if loop? Thx Colin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Performance issues
We actually use JSTL to support multiple locales and languages, so each web page calls its i18n taglib as many times as the number of text labels, currencies, dates, and other elements on the screen. JSTL slowdowns web server, we monitor an average 3-10% CPU comparing to 0,3-1% without JSTL (in the same application). I think overload of JSTL can reduce by 2-3 times the number of web application on a dedicated server to run at resonable performance. Evgeny Gesin Javadesk CEO / Founder --- Colin Chalmers [EMAIL PROTECTED] wrote: Karr, David wrote: Just in case, are you doing your measurements AFTER the first generation/compile of the servlet? It's not meaningful to measure taglib performance before then. Measurements are taking place during a continuous load of approx 60 users. With respect to caching evaluation results, it's probably not worth it, as it would have to do most of the evaluation before it could determine that nothing has changed, causing a cache read. OK, thx for the reply Colin -Original Message- From: Colin Chalmers [mailto:[EMAIL PROTECTED] Hi all, We've just started implementing TagLibs in our App and are seeing a couple of performance issues which I'd like to check with the group. We make heavy use of theExpressionEvaluationManger, is there any way to tweak this so that it pools/caches results? We also make a lot of use of the choose/when/otherwise Can this be a potential bottleneck compared to using a simple if loop? Thx Colin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] __ Do you Yahoo!? Yahoo! Hotjobs: Enter the Signing Bonus Sweepstakes http://hotjobs.sweepstakes.yahoo.com/signingbonus - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JSTL with Resin 2.1.11 and JDK1.4.2
[EMAIL PROTECTED] wrote: Actually, that is not true. According to the Resin 2.1.12 docs: --- servlet-classloader-hack Enables the Servlet specification classloader hack. The Servlet 2.3 classloader order violates the JDK classloader specification. By default, therefore, Resin follows the JDK requirements. Those application which need to violate the JDK spec may enable the servlet-classloader-hack. By default, the hack is disabled. --- I don't believe the behavior is different in Resin 3.0.x, but the configuration has changed. I'm using 3.0.x and it's loading my classes from WEB-INF/lib just fine. Does it only occur on classes already loaded by the JDK? Patrick - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Performance issues
We are also concerning i18n implementation impact on server performance. The overhead you mentioned is a logical result of the i18n approach you use. If there is not the fmt tag in a JSP file, the file will be cached in the memory after it is translated. That is a behaviour in Tomcat, may be the same for other containers. To have dynamic content of a JSP file, a set of messages need to be pulled into the file when a locale new request comes in. That is how the overhead occurs. To resolve this performance, you shall use the other approach: a set of JSP files per locale. This approach requests almost double development work on JSP files unless there were a localization JSP file editor. -- - Original Message - DATE: Sun, 18 Jan 2004 05:34:15 From: Evgeny Gesin [EMAIL PROTECTED] To: Tag Libraries Users List [EMAIL PROTECTED] Cc: We actually use JSTL to support multiple locales and languages, so each web page calls its i18n taglib as many times as the number of text labels, currencies, dates, and other elements on the screen. JSTL slowdowns web server, we monitor an average 3-10% CPU comparing to 0,3-1% without JSTL (in the same application). I think overload of JSTL can reduce by 2-3 times the number of web application on a dedicated server to run at resonable performance. Evgeny Gesin Javadesk CEO / Founder --- Colin Chalmers [EMAIL PROTECTED] wrote: Karr, David wrote: Just in case, are you doing your measurements AFTER the first generation/compile of the servlet? It's not meaningful to measure taglib performance before then. Measurements are taking place during a continuous load of approx 60 users. With respect to caching evaluation results, it's probably not worth it, as it would have to do most of the evaluation before it could determine that nothing has changed, causing a cache read. OK, thx for the reply Colin -Original Message- From: Colin Chalmers [mailto:[EMAIL PROTECTED] Hi all, We've just started implementing TagLibs in our App and are seeing a couple of performance issues which I'd like to check with the group. We make heavy use of theExpressionEvaluationManger, is there any way to tweak this so that it pools/caches results? We also make a lot of use of the choose/when/otherwise Can this be a potential bottleneck compared to using a simple if loop? Thx Colin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] __ Do you Yahoo!? Yahoo! Hotjobs: Enter the Signing Bonus Sweepstakes http://hotjobs.sweepstakes.yahoo.com/signingbonus - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] Get advanced SPAM filtering on Webmail or POP Mail ... Get Lycos Mail! http://login.mail.lycos.com/r/referral?aid=27005 - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Performance issues
To resolve this performance, you shall use the other approach: a set of JSP files per locale. This approach requests almost double development work on JSP files unless there were a localization JSP file editor. I want to avoid development of the same JSP for each locale to reduce the cost of project. Also future support will cost less in case of any change/adds. It is easer and more cost-effective to add more hardware/servers. Evgeny Gesin Javadesk CEO / Founder __ Do you Yahoo!? Yahoo! Hotjobs: Enter the Signing Bonus Sweepstakes http://hotjobs.sweepstakes.yahoo.com/signingbonus - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: Performance issues
My point is, it would be good to make sure your measurements don't include the initial generation and compile of the servlet. This will take quite a while, relatively, and will only happen once. Depending on how much continuous load you're providing, it may be less of a factor, but you should be aware of that, in any case. -Original Message- From: Colin Chalmers [mailto:[EMAIL PROTECTED] Sent: Saturday, January 17, 2004 5:17 AM To: Tag Libraries Users List Subject: Re: Performance issues Karr, David wrote: Just in case, are you doing your measurements AFTER the first generation/compile of the servlet? It's not meaningful to measure taglib performance before then. Measurements are taking place during a continuous load of approx 60 users. With respect to caching evaluation results, it's probably not worth it, as it would have to do most of the evaluation before it could determine that nothing has changed, causing a cache read. OK, thx for the reply Colin -Original Message- From: Colin Chalmers [mailto:[EMAIL PROTECTED] Hi all, We've just started implementing TagLibs in our App and are seeing a couple of performance issues which I'd like to check with the group. We make heavy use of theExpressionEvaluationManger, is there any way to tweak this so that it pools/caches results? We also make a lot of use of the choose/when/otherwise Can this be a potential bottleneck compared to using a simple if loop? Thx Colin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Performance issues
To resolve this performance, you shall use the other approach: a set of JSP files per locale. This approach requests almost double development work on JSP files unless there were a localization JSP file editor. I want to avoid development of the same JSP for each locale to reduce the cost of project. Also future support will cost less in case of any change/adds. It is easer and more cost-effective to add more hardware/servers. In fact, it is not whole lot more work to create multiple language versions of the same JSP file. Copy/paste and change the text/unicode is all the task. The tedious part is editing the existing JSP files. IBM has a resource bundle property editor. If it is a ongoing long term issue in your company, you might consider to build such tool for JSP. The resource bundle approach doesn't work well if a JSP file contains a big portion of text. There are some other similar siutations such as tooltip. It is not a good practice of using a condition statement in JSP for involving locale related content. If you use Tiles, a layout management component, you would find that you have not other choice, but use a-set-of-JSP-per-locale. Get advanced SPAM filtering on Webmail or POP Mail ... Get Lycos Mail! http://login.mail.lycos.com/r/referral?aid=27005 - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Performance issues
--- Vernon Smith [EMAIL PROTECTED] wrote: In fact, it is not whole lot more work to create multiple language versions of the same JSP file. Copy/paste and change the text/unicode is all the task. Technical writer, not a programmer, can write text in a resource bundle file in other language. The tedious part is editing the existing JSP files. That's the main reason to go with resource bundle. The resource bundle approach doesn't work well if a JSP file contains a big portion of text. I have few FAQs and some other docs, which completely translated in other language without using resource bundle. It is not a good practice of using a condition statement in JSP for involving locale related content. Agree, but this can be easily avoided by using parameters in resource bundle (think about MessageFormat) If you use Tiles, a layout management component, you would find that you have not other choice, but use a-set-of-JSP-per-locale. I use in-house developed taglib, similar to Tiles, and I have no problem using resource bundle approach. I think Tiles should also allow it. To consider when using resource bundle: 1. Performance: this should be tested, tested.. 2. Naming convention to better manage, from programmers point of view, a high number of JSTL tags in JSPs. Evgeny Gesin Javadesk CEO / Founder __ Do you Yahoo!? Yahoo! Hotjobs: Enter the Signing Bonus Sweepstakes http://hotjobs.sweepstakes.yahoo.com/signingbonus - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
SQL / Date query
I'm trying to create a JSP page the will display rows of a table beginning on the current date and ending at a total of 7 days. The following query works in MySQL but returns the following error when attempting to run the jsp: org.apache.jasper.JasperException: SELECT date, time, am_pm, height, cond FROM cherry_point_tides WHERE date BETWEEN CURDATE and CURDATE() + INTERVAL 6 DAY : Column not found, message from server: Unknown column 'CURDATE' in 'where clause' at org.apache.jasper.servlet.JspServletWrapper.service(JspServletWrapper.java:254) at org.apache.jasper.servlet.JspServlet.serviceJspFile(JspServlet.java:295) at org.apache.jasper.servlet.JspServlet.service(JspServlet.java:241) at javax.servlet.http.HttpServlet.service(HttpServlet.java:853) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: SQL / Date query
Hi Jack, Looks like you forgot to user parentesis in the first CURDATE. Your WHERE statement should be: WHERE date BETWEEN CURDATE() and CURDATE() + INTERVAL 6 DAY Regards, Felipe On Sunday 18 January 2004 22:48, Jack Lauman wrote: I'm trying to create a JSP page the will display rows of a table beginning on the current date and ending at a total of 7 days. The following query works in MySQL but returns the following error when attempting to run the jsp: org.apache.jasper.JasperException: SELECT date, time, am_pm, height, cond FROM cherry_point_tides WHERE date BETWEEN CURDATE and CURDATE() + INTERVAL 6 DAY : Column not found, message from server: Unknown column 'CURDATE' in - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]