Re: HSSF Serializer Questions
Andrew C. Oliver wrote: Danny Mui wrote: I'm currently in the process of enhancing the serializer to support more features (one pending patch in bugzilla for gnumeric merged regions..) and was wondering one thing. Is there a reason why the latest 1.9 dev snapshot is not being used as the target? There's a minor API break (getFormat -> getBuiltinFormat for 1.9) but everything seems to work in 1.9 as it did in 1.7. I haven't gotten around to it. I was waiting to see what happened with Morphos but I think that effort is effectively suffering from the jakarta-commons effect, so the show should go on. Proceed. Yes, please do. I'm the main Morphos developer, and my intention is to bring it to Cocoon. For those that have seen Morphos, this is the idea: 1) use the Cocoon pipelines as MorpherPipelines 2) use Cocoon blocks as "Morphers" This means that Morphos could easily become a simple frontend to Cocoon pipelines without the sitemap, to be used programmatically. Bottom line: please end in stuff :-) -- Nicola Ken Barozzi [EMAIL PROTECTED] - verba volant, scripta manent - (discussions get forgotten, just code remains) - - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
Re: HSSF Serializer Questions
Danny Mui wrote: I'm currently in the process of enhancing the serializer to support more features (one pending patch in bugzilla for gnumeric merged regions..) and was wondering one thing. Is there a reason why the latest 1.9 dev snapshot is not being used as the target? There's a minor API break (getFormat -> getBuiltinFormat for 1.9) but everything seems to work in 1.9 as it did in 1.7. I haven't gotten around to it. I was waiting to see what happened with Morphos but I think that effort is effectively suffering from the jakarta-commons effect, so the show should go on. Proceed. My stepsons are going to their dad's for christmas so I'll have more time soon. There should be relatively low latency on patches, but you might want to put me in the cc: because then they get sorted in my "to me" box... Another question, when gnumeric generates a lot of styles to interpret the fonts/styles, a fontrecord is added for each and every style/font. Excel '97 Complains it is out of memory with a filesize of 133kb, so i tried to reuse FontRecords when appropriate. Steps: Yes. 1. Created a duplicate object called Font that is really just HSSFFont with a public constructor 2. Created a toString representation of the font record so it can be added to a hashmap (basically concatenates all the properties) 3. Added another method to the serializer representation of the workbook (createFont(Font)) so that fonts aren't created all the time and reused. Does this method look correct (algorithm/approach wise)? I'd like to do this for styles as well since this approach cut the filesize down to 103kb. Well you kinda want the fonts to match to the style regions in theory. But it sounds right. -Andy Thanks! danny - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
HSSF Serializer Questions
I'm currently in the process of enhancing the serializer to support more features (one pending patch in bugzilla for gnumeric merged regions..) and was wondering one thing. Is there a reason why the latest 1.9 dev snapshot is not being used as the target? There's a minor API break (getFormat -> getBuiltinFormat for 1.9) but everything seems to work in 1.9 as it did in 1.7. Another question, when gnumeric generates a lot of styles to interpret the fonts/styles, a fontrecord is added for each and every style/font. Excel '97 Complains it is out of memory with a filesize of 133kb, so i tried to reuse FontRecords when appropriate. Steps: 1. Created a duplicate object called Font that is really just HSSFFont with a public constructor 2. Created a toString representation of the font record so it can be added to a hashmap (basically concatenates all the properties) 3. Added another method to the serializer representation of the workbook (createFont(Font)) so that fonts aren't created all the time and reused. Does this method look correct (algorithm/approach wise)? I'd like to do this for styles as well since this approach cut the filesize down to 103kb. Thanks! danny - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
