Camelia, the Perl 6 bug

IRC log for #padre, 2010-07-21

| Channels | #padre index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:16 teejay joined #padre
00:52 Alias joined #padre
01:23 waxhead Alias, interesting blog post... always fascinating to read how systems work...
01:23 waxhead that's a big code base..
01:24 Alias It's a beast
01:24 Alias And it's 15 years old
01:25 Alias Project started in 1995
01:25 Alias It's been rewritten from scratch twice since then
01:25 Alias It has it's own ORM that predates Class::DBI, and an MVC system that predates Catalyst
01:25 Alias The svn history is almost 10 years old
01:26 Alias Nobody in the company was around for the earliest commits
01:26 Alias So it's a real look at ancient history in computing term
01:42 Alias I'll hopefully post a few more "How we do it" posts from time to time
01:42 Alias There's some stuff here that's better than the state of the art on the CPAN
01:42 Alias (Half of which I wrote, to toot my own horn) :)
02:10 waxhead nothing wrong with tooting your own horn.. :)
02:10 waxhead I guess that's why in part there's perl out there, the age of the system...
02:10 waxhead I wonder how many new projects get perl nod over say .Net
02:11 waxhead so do you guys do much SOA within your infrastructure?
02:58 cognominal joined #padre
03:10 Alias Yes, but not really by that name
03:10 Alias We have a middleware platform, that is basically in charge of letting everything talk to everything else
03:11 Alias It kind of acts as a sort of SOA platform
03:11 Alias And the new SAP platform has it's own set
03:11 Alias But we're certainly not doing it in a full blown IBM style SOAP + Enterprise Directory type manner
03:11 Alias Most of the interfaces themselves are done ad-hoc
03:12 Alias Some are json, some are REST with non-schema tag soup XML, and so on
03:12 Alias That said, there's 115 different customer intergrations, and so on
03:12 Alias So it's a bit like SOA, but without the uniformity that the Java/XML cabal would have you believe is needed :)
03:14 Alias And there's not currently any live calls INTO the Perl stack
03:14 Alias Which is going to change later this year
03:51 |Sno| joined #padre
03:59 [Sno] joined #padre
05:19 waxhead Alias, I'm not sure the SOA stuff is the be all and end all.  Granted UDDI WSDL and the rest have their places, the reality is the interchange is agreed between people, which means that information is generally described in a document of agreement between the two parties
05:20 waxhead Data exchange/interchange is important and should be done whereever it makes sense to do so, but I'm not convinced about SOA in it's 'full blown' IBM/Websphere "your company belongs to us" style
05:20 Alias In some cases I can see how it makes sense
05:21 waxhead that seems to come at a fairly large cost though
05:21 waxhead both in terms of $$ and complexity...
05:21 Alias If, and only if, you have a massive amount of interfaces, and they are either rarely used or are completely elastically performant
05:21 waxhead at the end of the complexity will end up somewhere, you can't abstract it away to invisibility...
05:21 Alias I've seen some neat stuff though
05:22 Alias Where you get EXCELLENT IDE GUI magic support from the UDDI approach
05:22 Alias The interfaces are well-defined enough to let your IDE bring up input fields and output fields and compile-validate the types and what not
05:22 Alias What worries me though, is that you make interface $foo and put it into the catalog
05:23 Alias And then some other department likes the look of it
05:23 Alias So they start mass-calling it 100 times more often then you had expected
05:23 Alias And that service falls over
05:23 waxhead sure.. so you save some programmer some effort on writing their own class...
05:23 Alias So the extra rigour makes sense to me when the scale is large enough, hundreds or thousands of services
05:23 waxhead wouldn't that imply the wrong interface then?
05:24 Alias But in that situation, I don't understand how you do performance and load control
05:24 Alias I have this problem today
05:24 Alias Some types of load are real-time or user-critical
05:24 Alias A slow down in process X directly steals time from an expensive human
05:24 waxhead hence, the dept should have engaged in the right area to specify their needs and from there the SME's return an appropriate mechanism
05:25 Alias Other types of load are transactional and can be done as slowly as you want, as long as total throughput is high enough
05:25 waxhead be that direct connects to the database via Stored Procs or views
05:25 Alias Well, whatever
05:25 Alias but the UDDI catalog shortcuts the whole process
05:25 waxhead or SOA, being SOAP calls etc
05:25 Alias It's like saying "Here's an interface, go nuts"
05:25 waxhead ha.. yep
05:25 waxhead but then given that, you can protect the information to only those you give visibility to
05:26 Alias The middleware guys here put a lot of attention to the modality of each interface
05:26 waxhead which would mean that the dept should have contacted you to discuss their requirements
05:26 Alias They are valuable because they can adapt to give both parties the interface style they want
05:26 waxhead that's it..
05:26 Alias And provide either realtime, or fire-and-forget, or broken-stored, etc etc
05:26 Alias They do throttling, and load balancing etc
05:26 Alias If you need it
05:27 Alias The platform itself contains no state at all, no database and no data is stored in it
05:27 Alias Everything is a transaction
05:27 Alias It is there only to solve the real world problems around interfacing, without having to concern themselves about the actual semantics
05:28 Alias Except to do format translation or whatever
05:28 Alias And thus simplify all the programs talking to it, and make them more maintainable
05:28 waxhead so where do you sit in the scheme of things?
05:28 Alias Organisationally, I'm the "Architect" for the Perl stack
05:28 waxhead I'd have thought that based on your blog you're in the middle with the perl stuff
05:29 Alias I'm also loaned out to other teams
05:29 waxhead Who are the doers for the perl stack then?
05:29 Alias As a kind of devil's advocate
05:29 Alias There's 7 on the team
05:29 waxhead that's a fun job
05:30 waxhead 7 perl programmers?
05:30 Alias One Team Leader/Manager, one architect, 5 line coders
05:30 Alias So apart from the non-coding part of my job, I do a lot of deep hacking
05:30 Alias I'm finishing up a rewrite of Aspect.pm at the moment
05:31 Alias For our new telemetry system, which will produce metrics to support our push for usability
05:31 waxhead yeah, I saw that you were doing that too
05:31 Alias We need a kind of Perl Oscillascope
05:31 Alias To hook temporarily into the program running on the production cluster
05:31 Alias And spew analysis and metrics out to the main syslog server and log search engine
05:32 waxhead so is the web site done in perl too?
05:32 Alias Yup
05:32 Alias It's only the website
05:32 waxhead 250k lines for the web site??
05:32 Alias There's two ERPs, an old on in an archaic system called Universal, and a new on in SAP
05:32 Alias 250k lines for a website, yes
05:33 Alias It's a big website, you have no idea how complicated some of the stuff we do it
05:33 Alias is
05:33 waxhead clearly...
05:33 Alias Every one of the 100,000 users gets a different catalog
05:33 Hyppolit svn: r11913 | szabgab++ | http://padre.perlide.org/trac/changeset/11913
05:33 Hyppolit eliminate debugging print to console
05:33 Hyppolit trunk/Padre-Plugin-Perl6/lib/Padre/Plugin/Perl6/
05:33 Alias Every one of the 100,000 users can get different prices
05:33 Hyppolit svn: r11914 | szabgab++ | http://padre.perlide.org/trac/changeset/11914
05:33 Hyppolit remove the generated mo files from SpellChecker
05:33 Hyppolit trunk/Padre-Plugin-SpellCheck/share/locale/
05:33 Hyppolit svn: r11915 | szabgab++ | http://padre.perlide.org/trac/changeset/11915
05:33 Hyppolit giignore some generated .mo files
05:33 Hyppolit trunk/
05:33 Alias We support multiple different manager approval schemes
05:33 Hyppolit svn: r11916 | szabgab++ | http://padre.perlide.org/trac/changeset/11916
05:33 Hyppolit better error message in tests
05:33 Hyppolit trunk/Padre/xt/
05:33 waxhead oh... so a form of workflow too?
05:34 Alias And... and... and...
05:34 Alias You can build your own business cards
05:34 waxhead yep.. fair enough...
05:34 waxhead seriously?
05:34 Alias Based on a procedurally-defined corporate template provided by your marketing department
05:34 waxhead so a sort of drag and drp thingy?
05:34 waxhead oh...
05:34 Alias And then we can drop ship that design to a network of print shops around the country
05:34 kanishka joined #padre
05:34 waxhead cool...
05:34 Alias The cards redesign themselves on the fly depending on what fields you want to show on your card
05:35 waxhead done in perl?
05:35 Alias yyup
05:35 waxhead so you have a list of fields to add/remove?
05:35 Hyppolit svn: r11917 | szabgab++ | http://padre.perlide.org/trac/changeset/11917
05:35 Hyppolit svn:ignore some generated .mo files
05:35 Hyppolit trunk/Padre-Plugin-LaTeX/share/locale/ trunk/Padre-Plugin-SpellCheck/share/locale/
05:35 waxhead from that the template is done?
05:35 Alias And we can show you a digital proof on screen that we can PROVE is identical to the physical product that will show up on your desk
05:35 Alias Kerning bugs and all
05:35 waxhead wow.. how do you do that?
05:35 Alias So you never have to send physical document print proofs
05:35 waxhead svg?
05:36 Alias PDF and Postscript
05:36 waxhead interesting...
05:36 waxhead what do you use for PDF?
05:36 waxhead I've played with the PDF stuff in the past...
05:36 Alias Our own deep template object model, dumped out to PDF::API2, then to two different versions of ghostscript
05:37 Alias We support ordering of alchohol online
05:37 waxhead so you used PDF::API2... I think that was the one I last tried
05:37 Alias So we track responsible receiving party, and 18+ checks
05:37 Alias We do integration with numbering schemes of customers
05:37 Alias So you can order products using your own private product codes
05:38 Alias We do stupid amounts of stuff
05:38 Alias It all adds up
05:38 Alias Plus lots of security stuff, and audit trails, and so on
05:38 Alias Our main database has 180 tables
05:38 Alias This is Just the website
05:38 Alias We then send complete orders on to the "real" system
05:39 Alias And every feature has to work with every other feature
05:39 Alias For example, the search engine knows the products you are allowed to order and only returns those
05:40 Alias If you type your own name into the main search, it will find the business card with your name on it your secretary ordered last year
05:40 Alias The edge cases alone are insane
05:41 waxhead wow... complex.. :(
05:41 waxhead but awesome too..
05:41 waxhead I like stuff like that...
05:41 Alias If your marketing department decides to change your business card, we have to find all the orders on hold pending management approval, unwind them back into incomplete orders, redesign your card automatically to the new template, then mail you to tell you to reproof the card to make sure it is ok
05:42 Alias But if the redesign is trivially minor, we will subtly redesign the card WITHOUT telling you, and let the order continue normally
05:43 Alias If one person does all the ordering for 10 different departments, we have to support that, letting people move orders to different cost centres
05:43 Alias And those different cost centres may have access to different products, or get different prices
05:43 Alias We let you embed the entire website as a widget inside your giant corporate ERP system
05:44 waxhead where is all the querying done?  in the code or through stored procs/views?
05:44 Alias A mix
05:45 Alias Sometimes the Perl code needs to know about things the user isn't allowed to see, other times we don't WANT it to know it
05:46 Alias For example, if a manager for a person is in a different company
05:46 Alias The manager can't order product $foo, but needs to see it only while approving someone else's order
05:46 Alias Edge cases are the bane of our existance
05:46 Alias Hence stuff like this oscilliscope
05:47 Alias The real world is too complex for us to predict it, or cover all the edge cases
05:47 Alias So we want to say "On production, does this function ever REALLY get called with value X set to false?"
05:47 Alias So turn on a hook, and leave it on for three months and see
05:47 Alias And do so relatively cheaply
05:47 Alias Thus, Aspect
05:48 Alias Most time I think I've ever spent on company time working on a CPAN module
05:51 waxhead I remember reading your blog post about Aspect... it's a pretty cool thing to be able to do..
05:52 Alias It's one of those things that has a massive potential capability, but most of those things are fairly evil
05:52 Alias And only 10% of it's potential is actually worth using
05:52 Alias its
06:34 Haaarg joined #padre
07:06 daxim joined #padre
07:38 danlucraft joined #padre
07:40 danlucraft1 joined #padre
07:43 toi joined #padre
07:47 rindolf joined #padre
08:24 marcela joined #padre
09:21 zenog Hi guys.
09:23 Steffen joined #padre
09:29 zenog Hi Steffen
09:44 zenog I have a problem with the plugin translations (a) from a developer's perspective, and (b) from a user's perspective.
09:44 zenog The documentation on how to properly localize plugins is not so clear.
09:44 zenog Plus (b) none of the plugins I have installed display translated strings.
09:44 zenog ... even though they have .po files for German.
09:45 zenog This happens both for plugins I installed from CPAN (PerlTidy) as well as for plugins I directly installed from the SVN repository (e.g. Catalyst, SpellCheck)
09:46 zenog Anyone got an idea? Is this a known problem?
09:48 Hyppolit #741: Expected language settings in the Preferences, not under "View" (new task) [ http://padre.perlide.org/trac/ticket/741#comment:1 ]
09:52 zenog I added ticket 741 as a rather easy issue to OpenHatch: http://openhatch.org/+projects/Padre
10:01 waxhead hey zenog.... no idea about the translation stuff.. sorry
10:01 zenog One question about the ticket system in trac: What is meant by a "task" vs. an "enhancement" or a "defect"? The difference between "defect" and "enhancement" is clear to me ...
10:02 zenog waxhead: No problem, but can you try whether in your installation there is the same problem?
10:02 zenog waxhead: Just switch to Spanish or German for second and see whether the plugins are translated ...
10:02 zenog waxhead: (just if you have time ...)
10:02 waxhead ok..
10:06 waxhead actually when I changed to spanish I lost all the menu entries fort he plugins I have enabled.
10:06 waxhead going back to english didn't bring them back, but I can see the cataylst panel in the bottom pane, so it would appear they are still loaded.
10:07 waxhead and I get this in the console: 20:03:39: Error: Cannot set locale to 'es_ES'.
10:07 zenog I see the same behavior.
10:07 zenog Plug-ins seem to remain loaded (according to plugin manager), but somehow the menu entries are lost.
10:08 zenog Looks like a bug, doesn't it?
10:09 waxhead yeah, or the interface hasn't been done to handle locale changes for plugins maybe...
10:09 waxhead which is still a bug
10:10 * waxhead is wondering what he was looking at before that handles locale changes....
10:11 Haaarg joined #padre
10:12 waxhead I wonder if it's a function of the refresh that it does the relocale stuff
10:12 zenog possibly, I have never had a look at that part of the code ...
10:14 zenog I have lunch now, will be back later.
10:14 waxhead zenog, Main.pm has this:
10:14 waxhead # Relocale the plug-ins
10:14 waxhead $self->ide->plugin_manager->relocale;
10:21 Jagdwurst joined #padre
10:39 waxhead zenog, are you testing with ./dev.pl or the released version from cpan?
10:43 waxhead hmm same problem with the release though
10:59 zenog ./dev.pl
11:01 zenog But it is the same for me when I run with the installed "padre".
11:01 zenog (both trunk)
11:03 waxhead ok...
11:04 waxhead it's interesting I've put a few print statements in and for catalyst I don't set the "bottom" tab get translated, even though the syntax and output tabs do
11:16 zenog Syntax/Output are not part of the Catalyst plugin, aren't they?
11:37 zenog Shall I create a ticket?
12:00 waxhead yeah, I reckon... I've been poking about with it and I think I'm starting to see where things go.. not 100% sure though.. will check again tomorrow if I get a chance...
12:00 waxhead a ticket would at least then mean it's recorded for someone to check into...
12:10 [Sno] joined #padre
13:11 zenog Does anyone in the channel see translated plugin messages?
13:13 zenog e.g. with the Catalyst or the PerlTidy plugin, or any other plugin outside core Padre?
13:25 Haaarg joined #padre
13:35 pece joined #padre
13:36 Hyppolit #970: Switching language removes plugin menus (new defect) [ http://padre.perlide.org/trac/ticket/970 ]
13:36 zenog Who will create ticket #1000?
13:36 zenog 30 to go ...
13:44 rindolf joined #padre
13:46 Getty i always find ticket 1 interesting, sadly this one was wasted here ;)
13:47 Getty http://dev.laptop.org/ticket/1
13:49 Hyppolit #318: The search should highlight all the hits with yellow marker (new enhancement) [ http://padre.perlide.org/trac/ticket/318#comment:1 ]
13:50 Hyppolit #319: highlight # TODO (new enhancement) [ http://padre.perlide.org/trac/ticket/319#comment:1 ]
13:53 siegfri3d joined #padre
13:56 Hyppolit #356: encoding problem (new defect) [ http://padre.perlide.org/trac/ticket/356#comment:1 ]
13:56 zenog Getty: Oh, maybe we could manually mess with the ticket system ... or you think about a nice ticket #1000 - it does not have to be #1 all of the time ;-)
13:57 zenog Getty: Do you have an instance of Padre opened? If so, could you check for a moment whether you get translated plugin messages (see above)?
14:02 Hyppolit #338: regex replace treats \n as 2 chars (new enhancement) [ http://padre.perlide.org/trac/ticket/338#comment:1 ]
14:03 Hyppolit #342: * No 'replace all' available that acts on ALL open files (new enhancement) [ http://padre.perlide.org/trac/ticket/342#comment:1 ]
14:04 Getty zenog: its updated to 0.12?
14:04 Getty AH! wait
14:04 Getty you mean in padre?
14:04 Getty i dont have padre installed
14:04 * Getty runs
14:06 zenog Man, why don't you have Padre installed?
14:06 zenog What do we need to fix/implement to get you to install it?
14:07 Hyppolit #336: listing all regexes in replace dialog (new defect) [ http://padre.perlide.org/trac/ticket/336#comment:1 ]
14:07 Getty i just need to try it again to use, it was always that way then when i tried to use it on windows as regular IDE, the problems forced me to go back for being productive, sadly i cant invest that much time in my IDE or dont need to so far, but i will try to change that
14:08 Hyppolit #613: Copy Ultraedit's "file dissapeared" semantics (new enhancement) [ http://padre.perlide.org/trac/ticket/613#comment:1 ]
14:09 zenog Getty: Installation problems? Instability? Lack of features?
14:12 PerlJam joined #padre
14:18 Getty zenog: instability
14:18 Getty zenog: and speed of samba access
14:18 Getty (which is always a common IDE problem)
14:18 Getty (i must also remark: windows user here)
14:23 Getty suddenly it gets silent ;) hrhr
14:31 zenog Dunno, I use it only on Linux ... I have heard it is a bit less stable on Windows ...
14:32 Hyppolit #704: Segfault when closing documentation dialog (new defect) [ http://padre.perlide.org/trac/ticket/704#comment:1 ]
14:34 zenog Wasn't silent because of your answer, but because I was on the phone ;-)
14:36 Hyppolit #704: Segfault when closing documentation dialog (new defect) [ http://padre.perlide.org/trac/ticket/704#comment:2 ]
14:37 Hyppolit #782: warning when no locale found on Windows (new defect) [ http://padre.perlide.org/trac/ticket/782#comment:2 ]
14:38 Hyppolit #898: "Find in Files" dialog: buttons too small on Ubuntu 9.10 (new defect) [ http://padre.perlide.org/trac/ticket/898#comment:1 ]
14:45 Getty zenog: yeah it got its problem on windows
14:45 Getty zenog: but that is clear
14:47 mib_lu857i joined #padre
15:30 Hyppolit #870: Padre crashs when closing "Open Resource" with the 'x' button from window manager (new defect) [ http://padre.perlide.org/trac/ticket/870#comment:2 ]
15:53 * garu waves from FISL
15:55 garu Alias: I'm trying to install padre on a mac but even as I upgrade cpan it fails some tests on File::HomeDir - could you help me debug them?
15:59 Steffen Alias, ping? Would you consider a Perl::MinimumVersion release with my 5.12 changes from April?
16:00 shadowpaste "garu" at 217.168.150.38 pasted "File::HomeDir test error on mac os x 10.6" (103 lines) at http://paste.scsys.co.uk/46967
16:09 zenog garu: Alias is most likely not available because of the time zone difference ...
16:09 garu yeah, 1am on a wednesday, I knew it was a long shot :)
16:10 garu zenog: if you - or anyone else - has any clues on it, please take your best shot :)
16:13 zenog garu: The first fail looks like a trivial platform issue - the file is on Mac OS X just not at the same position as in a 'classical' Unix (just my guess, no idea at all of Macs)
16:14 zenog garu: '11_darwin' looks like trivial coding errors that do not show up on other platforms because 11_darwin may be platform-specific ...
16:15 zenog garu: Those are only wild guesses. Can you try to install the module anyway, and see what happens?
16:26 garu hmmm... I'm not so sure if that's a good idea, specially since I *think* it should work on darwin
16:27 garu I 'l try doing it as a local user with local::lib first, see what happens - I've been trying to install it as root so far
16:31 zenog garu: Good luck!!
16:42 garu zenog: thanks!
16:42 garu it's not the only one failing though
16:42 garu Wx itself is not installing properly either
16:45 Sewi joined #padre
16:45 garu anyone had any luck installing padre on mac?
16:48 kthakore garu: oh hai!
16:49 garu kthakore: heya!
16:49 kthakore garu: oh man ... major war with Blaizer today on tabs v.s. spaces
16:49 garu sigh
16:49 kthakore yeah
16:50 kthakore it is such a trivial thing
16:50 kthakore I mean we picked one
16:50 kthakore but now we have to change just for this guy
16:50 kthakore but I finnaly got him to do tests
16:50 kthakore woot!
16:50 kthakore he is coming along
16:50 kthakore Blazier++
16:51 garu :)
16:59 rindolf joined #padre
17:02 zenog kthakore: So what now, tabs or spaces?
17:02 kthakore zenog: spaces
17:12 zenog kthakore: hm.
17:13 zenog done for today &
17:16 kthakore zenog: ??
18:49 patspam joined #padre
19:35 mib_9xnt6q joined #padre
19:36 kyanardag joined #padre
19:43 toi joined #padre
20:16 Steffen joined #padre
20:20 mib_jlerqz joined #padre
21:24 mib_mpmmi5 joined #padre
21:56 mofino joined #padre
22:12 hmm joined #padre
22:23 mib_1t3rkw joined #padre
23:18 Alias <garu>Alias: I'm trying to install padre on a mac but even as I upgrade cpan it fails some tests on File::HomeDir - could you help me debug them?
23:18 Alias Shoot
23:18 Alias I'm taking a look at Perl::MinimumVersion
23:36 zamolxes joined #padre

| Channels | #padre index | Today | | Search | Google Search | Plain-Text | summary