Yep, according to Google it takes a while. I would check back tomorrow. I know mine took almost a day to appear on my inbox. On 2009/11/05, at 16:15, Faito wrote:
- Pedro Melo
from email
"Hi Jeff, just browsed your talk, pretty cool stuff. I would like to point out two things though: * you should use file: URLs for your CPAN url_list to your CPAN::Mini mirror: http://search.cpan.org/dist...... * please mention local::lib - no root required! :) Bye"
- Pedro Melo
"As I said, I don't have a meaningful example to work with. Notice that the 6 lines sample doesn't even include a 6 or 7 op. You can see how I build my 12M lines example file by reading the build_input_source.pl. With an input source like that: * my script gives the exact output like the stats_basic_optimized.pl; * on my laptop, at the end of the run, RSS is about 3Mb. So I can only assume that there is something in the formatting of the data that is not covered on that sample, and therefore, the regexps that I used don't match the proper fields."
- Pedro Melo
"Oh, and another thing: there is no $date at line 172 of my script: http://github.com/melo...... So you are probably running some other script. Best regards,"
- Pedro Melo
"Can you run the max_rate_lines.pl? Thats the only one that I care about, really... The extents part was not documented and the script will probably choke on that one. Without access to the real data I cannot improve the stats script further. But the max_line_rate.pl should give a upper bound on speed. Can you run it for a couple of seconds, hit ctrl-c and post the results of the final report? Thanks,"
- Pedro Melo
"Celestino, the comparison with wc was with wc -l of course. Also, my whole point was *not* using <STDIN> to read. See the wc_batch.pl at http://github.com/melo...... (hit the 'raw' link to download). On my laptop: $ gzcat nfsd.gz | time wc -l 12236390 17.69 real 13.77 user 0.51 sys $ gzcat nfsd.gz | time /usr/bin/perl ./wc_batch.pl 12236390 7.49 real 5.66 user 0.60 sys On Linux wc -l wins, so basically Mac OS X is very stupid."
- Pedro Melo
"Olás, Não percebo a tua resposta à "There should be no National ID card". Dizes não mas gostas do Cartão do Cidadão? Ou é o meu detector de sarcasmo que falhou redondamente? IMHO, ter um sistema de identificação nacional é importante, facilita muitos serviços que podes colocar depois em cima disso. Mas sim, a existir que seja único. Quem queixa de questões de privacidade, não deve ter pesquisado o seu nome no Google ainda... :)"
- Pedro Melo
"Olás, excelente pergunta, e o melhor que consigo responder é não sei :) A lib que uso para a aparte de XMPP não suporta FileTransfer ainda. Acho que está uma pessoa a tratar disso, mas não tenho visto nada de novidades. Assim que tiver suporte, acrescento isso."
- Pedro Melo
"Hi David, I did read that page. But it seemed to me that most of the suggestions where Apache-based workarounds. Given that I'm trying not to use Apache, I was thinking on how to move a bit of the logic that Apache provides (DirectoryIndex and 404 generation) to the FastCGI handler. I'm having fun actually :)"
- Pedro Melo
"Yes, the () is required to implement constants, but I would say that it is a hack. Just because the prototype is () it doesn't mean that the output of the function is constant. As for ($), I just don't use it. I always write int(rand(5)). I guess it is the only way to write unary operators and for that it is useful. Maybe prototypes are the only way to do some things. I wonder if all the Moose sugar (has, extends, etc) would be possible without them,I don't know. But it is action at a distance and although I would be ok with that if properly documented (given that its not the Perl std behavior), in this case it was not documented at al. (BTW, ignore the 'edited by moderator' that appears on your comment. I hit the wrong button in the Disqus interface, sorry about that)"
- Pedro Melo
"Hi, Catalyst + Mason or Mojo + Mason is indeed an option. I'm just considering a pure Mason approach. Its an interesting exercise, given that I have a clean model to work with, the extra layer of controllers is not that important to me. The one-controller/multiple-views is also not a factor in this particular situation, so the question is really: do I need Catalyst or Mojo?"
- Pedro Melo
Pena ser complicado (mas não impossível) usá-lo para bater texto normalmente. Já tentei escrever alguns artigos com ele, mas o hard-break das linhas é uma chatice do caraças.
- Bruno Miguel
Não uso Emacs à muitos muitos anos. Passei para vim, e agora que estou em Mac OS X uso o TextMate. On 2009/08/30, at 17:15, Bruno Miguel wrote:
- Pedro Melo
from email
"HI larsbalker, thats an interesting approach. Actually this blog uses SSI to generate the right-hand side column on the article archives :) so I was aware of it. I particularly like the fact that (with nginx at least) SSI requests are executed in parallel, so it becomes very multi-core friendly. I should definitively consider it, and it is cache friendly (I can just pre-generate some of the boxes to disk). Thanks for the reminder."
- Pedro Melo
"Hi abraxxa, I've looked at Reaction a couple of times, and I read mst talk a bit about it. I *think* the last time I looked at it I finally understood where mst wants and how reaction works, but it seems to me to be a very step learning curve just now, so no, I don't plan on using it for this project. I think I need a smaller project to experiment with it first. And this one is not small at all."
- Pedro Melo
"Hi Aristotle, They way I do it, I can test each included box separately. The template calls a box API, that returns the box HTML, so I can test that box API with a normal, regular .t file. But yes, I don't like it either. And yes, that extra layer, making for a fatter view that eventually uses regular template files, thats what I'm looking for, which module should I use to program that part."
- Pedro Melo
"Hi hdp, thanks, I didn't know that you could run the CGI Mason under FastCGI. That will be helpful with some other projects, 100% Mason, that I have. I stayed with Mason 1.05 for quite some time, until 1.24 or 1.25, due to the internal buffer changes, until they where reverted, due to the performance decrease, and I didn't re-read the docs when I upgraded. I'm sure I'm missing a lot of new stuff."
- Pedro Melo
"I used HTML::Mason extensively in the past but I admit that I always felt a good tool for simple projects, extremelly easy to get started with, but as a template system under Catalyst or other frameworks, a big bloated. But yes, I might need to read more about the current state of HTML::Mason. I remember that I found component methods very primitive, but it is a old recollection. Autohandlers and dhandlers where great but didn't play right with multiple root sources AFAIR. The current EVOLUI.COM system is HTML::Mason-based BTW. For standalone projects one thing that I missed was FastCGI support. If it had that, I would consider it for a lot of stuff that I do."
- Pedro Melo