>>The way warnings may be grouped is for example: syntactical, related >>to project structure, related to security, related to code safety (as >>opposed to security - a suspicious cast, an object used in a strange >>way etc.), related to code sloppiness - unused variables, function >>arguments, ureacheable code), deprecation warnings.
The grouping of problems in Falcon is done via subclassing. AbstractSemanticProblem and its subclasses are an example of that. -Darrell On Thu, Jan 1, 2015 at 5:43 AM, Left Right <olegsivo...@gmail.com> wrote: > Well, if you ask me how'd I like this to be, then, probably, I'd like > this to have hierarchical structure (maybe I'm biased by how C > compilers go about it). In other words, I'd like warnings to be > grouped (not necessary by severity). Whether to warn on assignment > inside condition is a matter of style, while warnings on duplicated > source file most likely hint at an error. > > The way warnings may be grouped is for example: syntactical, related > to project structure, related to security, related to code safety (as > opposed to security - a suspicious cast, an object used in a strange > way etc.), related to code sloppiness - unused variables, function > arguments, ureacheable code), deprecation warnings. > > Then there could be some groups, which contain sets of warnings from > different categories (such as pedantic, all, dangerous). > > Re' me being a committer - nope, not yet at least. I'm trying to > convince my manager to let me migrate my project to use Falcon. > Provided I'll succeed, I'll get time to work on the Falcon sources > during my office hours and I'll be able to contribute more than just > emails :) But this is far from being certain as of now. Which brings > me to a completely unrelated question: if I wanted to convince someone > to try Falcon, what would be the good argument to do so? > > Best, > > Oleg > > I would also like there to be a more easily recognizable names than > 1234 and 5678 :) > > On Wed, Dec 31, 2014 at 10:29 PM, Gordon Smith <gsmit...@hotmail.com> > wrote: > > > > > > > >> Could this be improved to have a better interface? > > > > Darrell, don't -error-problems, -warning-problems, and -ignore-problems > allow the problems to be specified either by fully-qualified class name or > by numeric problem code? And isn't the numeric problem code displayed along > with the problem message? > > > > Left Right, are you already a committer? If not, do you want to be one > so that you can make improvements? If an option like > -ignore-problems=1234,5678 works, do you think that's intuitive enough? > > > >> I get 1388 hits for org.apache.flex.compiler.problems > > > > That's because there are 1388 or so classes representing compiler > problems. > > > >> No one would think of this as being an easy way to find an offending > warning > > > > The way to search the source for the class representing a particular > problem is to search for part of the English message. But you have to be > careful not to search for something that is getting dynamically substituted > for a placeholder in the string. > > > >> I don't think most people would even go as far as looking into the > source code for ways to void a warning message. > > > > Numeric problem codes seem like the way to go, if we don't already > support them. > > > > - Gordon > > > >> Date: Wed, 31 Dec 2014 11:56:45 +0200 > >> Subject: Re: [FALCON] don't warn on assignment in while (condition) body > >> From: olegsivo...@gmail.com > >> To: dev@flex.apache.org > >> > >> Could this be improved to have a better interface?.. Grepping through > >> the code I get 1388 hits for org.apache.flex.compiler.problems in Java > >> files alone. No one would think of this as being an easy way to find > >> an offending warning... But I don't think most people would even go as > >> far as looking into the source code for ways to void a warning > >> message. > >> > >> For those interested in this particular warning, I assume it's this > >> one: org.apache.flex.compiler.problems.AssignmentInConditionalProblem > >> > >> Besides, there doesn't seem to be a way to specify this in the mxmlc > Ant task... > >> > >> On Wed, Dec 31, 2014 at 4:28 AM, Darrell Loverin > >> <darrell.love...@gmail.com> wrote: > >> > The falcon compiler and the (old) mxmlc compiler handle errors and > warnings > >> > differently. In the mxmlc compiler a message is always an > >> > error/warning/info message at creation. In falcon, messages have a > default > >> > severity but can be treated as an error, warning, or ignored. The > >> > configuration options to put a message into a severity category are > >> > -error-problems, -warning-problmes, and -ignore-problems. > >> > > >> > So to suppress a warning use -ignore-problems, passing the > fully-qualified > >> > problem class to ignore. > >> > For example: > >> >>mxmlc -ignore-problems > >> > org.apache.flex.compiler.problems.ANELibraryNotAllowedProblem > >> > > >> > will ignore all reported problems with class > ANELibraryNotAllowedProblem. > >> > The compiler will still report the problem it will just won't be > displayed. > >> > For more info see the ProblemSettingsFilter class. This class handles > the > >> > filtering and implements mxmlc options that ignore warnings. > >> > > >> > > >> > -Darrell > >> > > >> > On Tue, Dec 30, 2014 at 11:41 AM, Alex Harui <aha...@adobe.com> > wrote: > >> > > >> >> I don’t know for sure. Maybe Gordon or Darrell know if warning > >> >> suppression is supposed to work in Falcon. > >> >> > >> >> On 12/30/14, 1:35 AM, "Left Right" <olegsivo...@gmail.com> wrote: > >> >> > >> >> >I looked into mxmlc -help warnings but I don't see an option to void > >> >> >the warning issued on assignment inside while (and maybe other such > >> >> >places). Is there one, or it simply isn't implemented yet? > >> >> > > >> >> >Thanks! > >> >> > >> >> > > > > >
