On 13/02/2015 18:49, Ryan Scharer wrote: > Chris, > > I share your misgivings about magic, though if it exhibits well-documented > and predictable behavior I usually just shrug and go along with it. Sadly > that doesn't seem to be the case here.
It is documented but I'd agree it could be better documented. I'll set aside some time to step > through the Tomcat code to try to figure this out, Save yourself the effort and read this from line ~1085. http://svn.eu.apache.org/viewvc/tomcat/tc8.0.x/trunk/java/org/apache/catalina/startup/ContextConfig.java?view=annotate Mark > though in the meantime > my best bet may be to repackage the 3rd party jar minus its web-fragment. > > -R > > On Fri, Feb 13, 2015 at 1:18 PM, Christopher Schultz < > ch...@christopherschultz.net> wrote: > > Ryan, > > On 2/13/15 12:59 PM, Ryan Scharer wrote: >>>> I'm not sure if this is a bug or not, but I can't find any relevant >>>> information in the spec to suggest the behavior is expected. >>>> >>>> There's a web-fragment in my classpath that I'd like to skip. The >>>> only way to accomplish this that I know of is to put an >>>> <absolute-ordering> stanza in my web.xml and omit an <others/>. >>>> Though this has the desired effect of skipping the web-fragment I >>>> don't want, it has the very negative side effect that my >>>> ServletContainerInitializer doesn't get handed all instances of >>>> WebApplicationInitializer anymore, despite its @HandlesTypes >>>> annotation. If I add the <others/>, classpath scanning works fine, >>>> but the undesired web-fragment comes back. I've tested this in the >>>> latest 7.x and 8.x Tomcat releases. >>>> >>>> This begs two questions: >>>> >>>> 1. Why does specifying an <absolute-ordering> without an <others/> >>>> kill classpath scanning, or at least the part of Tomcat >>>> responsible for finding types specified by @HandlesTypes and giving >>>> it to interested parties? 2. Is there an alternate way to skip a >>>> web-fragment, short of ripping it out of the jar, which I really >>>> don't want to do? > > It's unclear to me why <absolute-ordering> affects JAR scanning... > absolute-ordering should affect only the processing of web-fragments. > The Tomcat documentation for absolute-ordering makes it sounds like > you have to mention a JAR name while the spec documentation makes it > seem like you need to use the <name> element from web-fragments that > are detected in JAR files. > > The whole thing is a can of worms, honestly. > > As for your inability to skip a web-fragment... that seems > straightforward to me: if you have <others/> specified, then > "everything else" will be processed in that order, including the > web-fragment you don't want. There does not seem to be a way to > blacklist a web-fragment short of completely removing it from your > project's libraries. > > But the fact that the lack of <others> causes Tomcat to fail to do JAR > scanning is surprising to me. I tend to prefer explicit configuration > over all this scanning-related magic so I'm afraid I don't have any > experience with this kind of thing. In fact, it's this kind of > foolishness that makes me stick with explicitly-specified > configurations instead of magic ones. Good luck figuring out how the > magic works / is supposed to work! > > -chris >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org >> For additional commands, e-mail: users-h...@tomcat.apache.org >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
