esme-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Vladimir Ivanov <lester.m...@gmail.com>
Subject Re: Move to latest Lift release for 1.2
Date Tue, 14 Dec 2010 01:22:21 GMT
Funny thing that I did recognize that this partial function deals with
immutable list and understand that new List will be returned while
suggesting new method (I explicitly pointed to this fact in my first mail in
current discussion). I thought that '+=' is a method on List while it was
shortcut - that was my first mistake. Then I decided that (after calling new
method ':+') new instance of List should be returned from partial function,
while it reassigns new instance of List to the same reference in previous
version of PF - that was the second.

But I agree with you, for someone (like me :s) ) who use to coding in
imperative style (like in Java), functional style is not so obvious.

2010/12/14 Vassil Dichev <vdichev@apache.org>

> You're right, the value was appended before, but I used the
> opportunity to improve it now.
>
> I've done this same mistake before, that's why I was quick to notice
> it. It takes a while for someone accustomed to an imperative language
> to learn that these operators are not setters but just produce a new
> data structure.
>
> Vassil
>
>
> On Tue, Dec 14, 2010 at 12:49 AM, Vladimir Ivanov <lester.mail@gmail.com>
> wrote:
> > Does it mean that ordering of poolId elements in list is not important?
> Elem
> > was appended to list before and it is prepended to list now.
> >
> > I was sure that += 'operator' adds elem to the end of List, so I decided
> > that appending was required for some reason.
> >
> > 2010/12/14 Vassil Dichev <vdichev@apache.org>
> >
> >> Hey all,
> >>
> >> API2 has actually nothing to do with it, when you add a pool messages
> >> sent to that pool will disappear, never to be seen. The pool actually
> >> appears functional after restart.
> >>
> >> The problem is in the modifications needed to make ESME compile. When
> >> adding to a list, we changed the "+=" operator with ":+". It does
> >> create a new list with an element appended, but doesn't update the
> >> initial list variable. In order to do that, you can use ":+=" or just
> >> list = list :+ elem
> >>
> >> This operator doesn't prepend though, it appends, which has horrific
> >> performance parameters for lists. These lists are expected to be
> >> fairly small, but nevertheless, let's not get into bad habits. Here
> >> it's better to use the "+:" operator, which will turn into "+:=" if we
> >> want to update the initial reference. An even better choice would be
> >> to use the "::" operator, which is specific to lists and is the best
> >> option if we know it's lists we're dealing with, not some generic
> >> collection. So the final operator I've committed is "::="
> >>
> >> Hopefully this operator mumbo-jumbo wasn't too long and boring :)
> >>
> >> Vassil
> >>
> >>
> >> On Mon, Dec 13, 2010 at 8:37 AM, Ethan Jewett <esjewett@gmail.com>
> wrote:
> >> > Ah, excellent, thanks for checking the API2 issue. I guess it's time
> >> > to dig into that code again...
> >> >
> >> > Ethan
> >> >
> >> > On Monday, December 13, 2010, Vassil Dichev <vdichev@apache.org>
> wrote:
> >> >> I fixed the parsing error- a change in the API caused an implicit
> >> >> conversion not to trigger and the string for the hashtag was not what
> >> >> was expected.
> >> >>
> >> >> The API2 problem is not a failure of the test- messages in a pool
> >> >> don't show in the timeline, will need more investigation to find out
> >> >> why.
> >> >>
> >> >> Vassil
> >> >>
> >> >>
> >> >> On Sun, Dec 12, 2010 at 8:50 PM, Ethan Jewett <esjewett@gmail.com>
> >> wrote:
> >> >>> I think deploying to stax should be fine. I didn't test much
> >> >>> functionality in the lift-2.2 branch, but I didn't see anything
> >> >>> broken. Deploying on Stax would make it easier to manually test,
> >> >>> especially since we do some real-world usage on there.
> >> >>>
> >> >>> Ethan
> >> >>>
> >> >>> On Sun, Dec 12, 2010 at 6:44 PM, Richard Hirsch <
> hirsch.dick@gmail.com>
> >> wrote:
> >> >>>> Should I do a test deployment tomorrow on stax? Or should I
wait
> until
> >> >>>> to test errors are corrected. I could deploy the 2.2 branch
on the
> >> >>>> second stax instance for a first test.
> >> >>>>
> >> >>>> D,
> >> >>>>
> >> >>>> On Sun, Dec 12, 2010 at 12:36 PM, Ethan Jewett <esjewett@gmail.com
> >
> >> wrote:
> >> >>>>> Ok, this is applied and everything seems to be running
on my
> machine
> >> >>>>> as well. As you mentioned, there are some tests failing,
so we'll
> >> need
> >> >>>>> to get those resolved before merging back into trunk. I
don't
> think
> >> >>>>> I'll have much time to look at them today. I suspect that
at least
> >> the
> >> >>>>> API2 issues are timing-related - some of the tests have
race
> >> >>>>> conditions in them where they sit around for 2 or 3 seconds
> waiting
> >> >>>>> for a message to post, and it's possible that Lift 2.2
handles
> things
> >> >>>>> a bit differently causing the wait to end before the post
> completes.
> >> >>>>> Vassil has a nice way of doing tests without explicit wait
times
> in
> >> >>>>> the Twitter API tests, but we haven't gotten around to
copying it
> >> over
> >> >>>>> to the API2 tests.
> >> >>>>>
> >> >>>>> I've also pulled the latest changes from trunk into the
branch, so
> >> >>>>> that should pick up the fix to the user.html file, as well
as the
> >> >>>>> changes to the user.html timeline I made yesterday. These
changes
> are
> >> >>>>> not well formated or well-tested but they seems to work.
> >> >>>>>
> >> >>>>> Cheers,
> >> >>>>> Ethan
> >> >>>>>
> >> >>>>> On Sun, Dec 12, 2010 at 11:58 AM, Ethan Jewett <
> esjewett@gmail.com>
> >> wrote:
> >> >>>>>> Hi Vladimir,
> >> >>>>>>
> >> >>>>>> That is from my commit yesterday :-( Not sure how it
snuck in
> there,
> >> >>>>>> but I'm fixing it now. I'll take a quick look at your
pom.xml
> patch
> >> >>>>>> for the logger dependency and get that applied as well.
> >> >>>>>>
> >> >>>>>> Ethan
> >> >>>>>>
> >> >>>>>> On Sun, Dec 12, 2010 at 9:10 AM, Vladimir Ivanov <
> >> lester.mail@gmail.com> wrote:
> >> >>>>>>> Hello, Vassil!
> >> >>>>>>>
> >> >>>>>>> You are right. I just wondered why it was placed
to license
> >> section.
> >> >>>>>>>
> >> >>>>>>> 2010/12/12 Vassil Dichev <vdichev@apache.org>
> >> >>>>>>>
> >> >>>>>>>> Sorry I'm late to the party, but thanks Vladimir
for your
> efforts!
> >> >>>>>>>>
> >> >>>>>>>> Normally the parser would be right, this is
not valid XML:
> >> >>>>>>>>
> >> >>>>>>>> <img <user:image/> alt="" width="50px"/>
> >> >>>>>>>>
> >> >>>>>>>> ...unless it's in a comment. Maybe something
has changed in the
> >> >>>>>>>> parser, but if you just delete this line, you
should be fine.
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>> On Sun, Dec 12, 2010 at 3:06 AM, Vladimir Ivanov
<
> >> lester.mail@gmail.com>
> >> >>>>>>>> wrote:
> >> >>>>>>>> > But now I discovered error while trying
to see user's
> profile:
> >> >>>>>>>> >
> >> >>>>>>>> > Error locating template /info_view/user.html.
Message: :1:6:
> '>'
> >> expected
> >> >>>>>>>> > instead of '<'
> >> >>>>>>>> > scala.xml.dtd.ValidationException: :1:6:
'>' expected instead
> of
> >> '<'
> >> >>>>>>>> >
> >> >>>>>>>>
> >>
> net.liftweb.util.PCDataXmlParser.reportSyntaxError(PCDataMarkupParser.scala:175)
> >> >>>>>>>> >
> >> >>>>>--
> >> >> Twitter: http://twitter.com/vdichev
> >> >> Blog: http://speaking-my-language.blogspot.com
> >> >>
> >> >
> >>
> >>
> >>
> >> --
> >> Twitter: http://twitter.com/vdichev
> >> Blog: http://speaking-my-language.blogspot.com
> >>
> >
> >
> >
> > --
> > Best Regards,
> > Vladimir Ivanov
> >
>
>
>
> --
> Twitter: http://twitter.com/vdichev
> Blog: http://speaking-my-language.blogspot.com
>



-- 
Best Regards,
Vladimir Ivanov

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message