Remy Maucherat wrote:
Filip Hanik (lists) wrote:
BTW, is it ok if I start my refatoring(s) in a separate branch based on
go ahead, refator. :-)
I'll "refator" then.
I'll redo logging as well, and define c-l categories (for webapps, for example) to be able to eventually remove Logger altogether.
Other things:
- attempt to redo a bit the deployer to be able to remove the CL code which is there to avoid JAR locking (and then use the optimized CL for the whole container); this is for faster startup
- switch to something other than digester to remove some commons deps (?) (I'm thinking about beanutils and collections); if we don't do that, I think we should do package renaming on these utilities to lower the likelihood of a user being locked into a specific release
I think renaming the classes, like they did for Xerces in j2se 1.5, would be the easiest way. But since we are "refactoring", we can also think of a pluggable SPI were the digester can be "plugged", as well as our own parsing mechanism. That would be usefull when Tomcat is embedded and it's container already have a parsing mechanism, as well as allowing people to develop their own parsing mechanism on top of Xerces (or avoid Xerces).
-- Jeanfrancois
- I'll try to avoid touching the sensitive stuff as much as possible (connector, anything related with the servlet API impl) to avoid regressions
(obviously the two branches will need to remain synced)
I think 5.0.x should be put in a tomcat_5_branch, while the refactoring is done in HEAD. Or I use a separate folder.
Rémy
--------------------------------------------------------------------- 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]
