Request for review: 8017611: Auto corrector for mistyped vm options

Dmitry Samersoff dmitry.samersoff at oracle.com
Wed Jun 26 02:49:05 PDT 2013


Bengt,

I don't have a link to webrev, but we *do need* a flag that
turns this behavior on and it have to be off by default in production
build.

It's sysadmin nightmare when a program auto correct mistypings in
startup scripts rather than abort immediately.

-Dmitry



On 2013-06-26 09:05, Bengt Rutisson wrote:
> 
> Tao,
> 
> This is great! Thanks for doing this! :)
> 
> I haven't looked closely at the code yet, so this is not a review, but I
> agree with Christian that we don't need a flag for the "fuzziness".
> Also, it seems like it would be pretty straight forward to write a
> couple of tests for this feature. Maybe just two simple JTreg tests, one
> that tests that a misspelled command line option gets the right
> suggestion and one correctly spelled option that does not cause a
> suggestion.
> 
> Thanks again for coming up with this!
> Bengt
> 
> 
> 
> On 6/25/13 8:33 PM, Tao Mao wrote:
>> OK, I'll move the flag to be local.
>>
>> Thanks.
>> Tao
>>
>> On 6/25/13 11:13 AM, Vladimir Kozlov wrote:
>>> Yes we need this! How many times I mistyped PrintMiscellaneous :( ?
>>>
>>> Why we need additional flag VMOptionsFuzzyMatchSimilarity (we have
>>> too many already)? Can you use a local constant?
>>>
>>> Where is whole Christian's mail?
>>>
>>> I redirected this RFR to hotspot-dev since it affects everyone.
>>>
>>> thanks,
>>> Vladimir
>>>
>>> On 6/25/13 11:01 AM, Tao Mao wrote:
>>>> Hi Chris,
>>>>
>>>> So you suggest the flag be "notproduct" or "develop"?
>>>>
>>>> Thanks.
>>>> Tao
>>>>
>>>> On 6/25/13 10:59 AM, Christian Thalinger wrote:
>>>>> I think we can do one step further to interpret the user's intent and
>>>>> suggest the correct VM options. Then i start to imagine that we have
>>>>> something like this
> 


-- 
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.


More information about the hotspot-dev mailing list