seaside February 2009

seaside@lists.squeakfoundation.org

66 participants
65 discussions

How to turn off Seaside toolbar in a new window?
by Sophie (itsme213) 18 Jul '10

18 Jul '10

html popupAnchor callback: [self renderNewWindow]; toolbar: false; with: 'No Toolbar'. But that turns out to control the Browser toolbar, not the Seaside one. Thanks - Sophie

3 3

[ANN] Phantasia: A CSS Framework
by Hernán Morales 27 Oct '09

27 Oct '09

Dear anyone interested in working with Smalltalk objects: Now you can write in Smalltalk to style pages with CSS. Instead of styling with ugly, boring and long Strings (where you must perform a continuous visual sweeping of the code to make the most trivial modification), you can create high-level super sexy CSS constructs, track and review your styles with the standard Smalltalk tools (XReferences), and share constructs and configurations easily. The framework includes a set of 40 fixed layout templates ready for use, and let you compose the basic CSS elements as objects. A note: If your department, office, ark, whatever, have graphical designers, they rarely would love to learn this new strange prolific ninja environment called Smalltalk just for writing CSS - besides, a lot of them are already in love with anything after the word "Adobe" -, but, for smalltalkers, by far we believe that's easier to work inside Smalltalk with objects than in a file-based way. Sincerely, Hernán PD: The I-almost-forget-it part of this mail: Juan Burella and me wrote Phantasia. Download it from http://www.squeaksource.com/Phantasia.html

9 10

SEO and Seaside
by Richard Eng 27 Apr '09

27 Apr '09

Is Seaside/Squeak SEO-friendly? My website isn't showing up in any search results on Google, Yahoo!, or MSN. Apparently, spiders have a hard time scanning my website. Perhaps because of the "dynamic" nature of the application? Thanks, Richard

11 13

seaside vertical apps?
by John M McIntosh 18 Mar '09

18 Mar '09

I was talking to someone today and they were asking if there are vertical enterprise apps available for Seaside? Mmm good question I said, so are there? Surely someone must be building turn key systems they want to sell to mid-size or large corporations. So web URLs at this time would be helpful -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

3 2

[CONFUSED]: WAKom, WAKomEncoded or WAKomEncoded 3.9
by Udo Schneider 07 Mar '09

07 Mar '09

All, I'm currently working on a Seaside App which will go "live" next monday. Currently I'm using Damiens 3.10 based sq3.10.2-7179web09.02.1 image. Yesterday I had an horror experience when a Mac Betatester told me that the whole app got screwed up when he entered a euro sign (€) in the app. All following responses are totally garbage. Other characters like german umlauts (or even the euro sign from a windows client!) where fine and I tested them before - although they appear as (UTF8?) giberisch in the image. This triggered some memory and I changed from WAKom to WAKomEncoded which obviously "solved" the problem. However this morning (after reading the class comments) I'm a bit confused about the whole issue and I have two questions: 1) What's the differnece between WAKom, WAKomEncoded and WAKomEncoded39? I read the class comments but do not really understand the issue. In addition they are dealing with 3.8 vs. 3.9 and I'm on 3.10... If somebody advises me to deploy on 3.9 I'll do - I just need a working configuration. 2) I assume that the browser side always uses UTF-8. But in which format does the image "see" the input? CU, Udo

4 6

Client Framework Supports Authentication
by Keith Hodges 06 Mar '09

06 Mar '09

New: WAAuthenticationHelper, when enabled http authentication is validated via the "Client" framework, whatever that is set-up to do. example usage: My main pier application, requires users to log in. I also have a second admin application, and admin users can also log in there. The WAAuthenticationHelper allows the login to occur via http authentication, thus protecting the admin app, and still using the same account details that are maintained in the user database. regards Keith ===== The Client Framework provides: 1. An architecture for sensibly separating your front-end seaside presentation from the backend data model 2. A place to put configuration attributes, or any data, error messages, etc on a per client basis 3. Selectable backends for mock/test operations The Client-Login package adds seaside components for logging in. The Client-PierLogin package adds pier command for logging in, emailing a login link on a forgotten password etc.

2 2

Create an SUUpdater without using the WARenderCanvas
by Lautaro Fernández 03 Mar '09

03 Mar '09

Hello list, I was wondering if it's possible to create an instance of SUUpdater without using the normal: (html updater) id: 'anID'; callback: [:renderer | someNiceRendering ]. If I do the following: updater := SUUpdater new. updater url: 'http://localhost:8080/seaside/meteoroid/counter'. " == html context actionUrl withoutParameters" updater addParameters: #('_s'->'GaMJovPCAHDxFKZT' '_k'->'VLnoRkpz'). " == html context actionUrl parameters" updater id: 'anID'; callback: [:renderer | someNiceRendering ]. It generates a JS(*), but when it renders the Update it does inside a div, not inside the body. (*) meteoroid.eval('new Ajax.Updater(\'count\',\' http://localhost:8080/seaside/meteoroid/counter \',{\'evalScripts\':true,\'parameters\':[\'_s=GaMJovPCAHDxFKZT\',\'_k=VLnoRkpz\'].join(\'&\')})') So, It's possible to create a SUUpdater without the normal html? (of course, it has to work in an page) Thanks in advance Lautaro Fernández PS: Why do I need this? because I'm doing a rendering process inside a model and I'm inserting that updater into the page via Comet, but I don't want to keep the 'html' inside the page as an instance variable, just the values I need (like the url and the parameters). -- Luke LAut SkyFernadezWalker

3 5

Re: How to profile a server image?
by Andreas Raab 01 Mar '09

01 Mar '09

[Apologies for cross-posting but I figured some of the more interested audience may be on the Seaside list. What follows below is a summary of my experiences with server profiling over the last couple of years which I've done a goodly bit of] Herbert König wrote: > say I have a web server with several users logged into different > sites. > > What do i have to start a MessageTally on to see where the image > spends its time? This is tricky with MessageTally. There are several issues to keep in mind, some of which you can fix, some of which you can't. Here we go: 1) Observing all Processes -------------------------- First, MessageTally only observes the block you are running, not the entire system. For example, if you were to profile this: done := Semaphore new. worker:= [ 10 timesRepeat:[World changed; displayWorld]. done signal. ] fork. MessageTally spyOn:[done wait]. This is what you'd get: **Tree** 100.0% {2367ms} primitives **Leaves** 100.0% {2367ms} UndefinedObject>>DoIt Obviously it's not actually measuring what is going on during the period of time which is a real problem if you are trying to measure server load in general. But it's fixable. 2) Priority issues ------------------ MessageTally uses userInterruptPriority and won't see anything running (for example) at lowIOPriority or higher. This is quite common for server images for handling incoming requests etc. The good news is, both of the above issues can be fixed easily. Just file in the attached patch for MessageTally (I will also post this on Mantis) and it will report the work being done accurately when you use it via: MessageTally spyAt: Processor timingPriority-1 on:[done wait]. The priority argument allows you to specify at which priority to run your spy so that you can measure what you need to measure (a word of warning though: You mustn't try to run at #timingPriority as this will interfere with the timing mechanism itself). With these changes you can run a tally on, e.g., MessageTally spyOn:[(Delay forSeconds: 10) wait]. and see what happens over the next 10 seconds. This is effectively what we do for profiling on our server images. 3) VM timing problems --------------------- Unfortunately, recent VMs have a problem with spy accuracy since an important timer check was removed from the VM (I didn't notice that at the time it happened). This *heavily* affects the profiler, for example, try this: MessageTally spyOn:[ 1 to: 100 do:[:i| Display fillWhite. 'Hello', ' World'. ]. ]. If you have a recent VM (I haven't tracked how far back this goes) the result is this: **Tree** 100.0% {534ms} ByteString(SequenceableCollection)>>, **Leaves** 100.0% {534ms} ByteString(SequenceableCollection)>>, The problem is caused by the VM no longer doing a timer check after a primitive has been run. Consequently, the time spent in any long-running primitive (like BitBlt) is allocated to the *next real send* when an interrupt check occurs (as you can see in the above). This can be extremely confusing. If you want even remotely accurate profiles I recommend going back a few generations in VMs and pick one that behaves properly in this regard. For example, running the above in 3.6 I get: **Tree** 100.0% {218ms} DisplayScreen(DisplayMedium)>>fillWhite 100.0% {218ms} DisplayScreen(DisplayMedium)>>fill:fillColor: 100.0% {218ms} DisplayScreen(Form)>>fill:rule:fillColor: 100.0% {218ms} BitBlt>>copy:from:in:fillColor:rule: **Leaves** 100.0% {218ms} BitBlt>>copy:from:in:fillColor:rule: Obviously, this is the correct tally. [Hm ... but this points out another issue. For some reason 3.6 runs this in half the total time??? What the heck... I'll have to look at this] You *must* use a VM that doesn't have the above problem or the results will be completely confusing. 4) Measuring load with (sub-)msecs responses -------------------------------------------- Lastly, server images can have a particular issue if they are set up to do relatively fast request handling. For example, our routers generally act in sub-msecs responses. When you start putting load on servers that react that quickly, your whole msec-based sampling approach goes out of the window. What's worse than that is that if (some of) your server activities are driven by the clock (using Delay etc) running under load can produce *utter* garbage result. Another example: blamedProcess := [ [true] whileTrue:[(Delay forMilliseconds: 1) wait]. ] forkAt: Processor userInterruptPriority. [MessageTally spyAt: Processor timingPriority -1 on:[ (Delay forSeconds: 5) wait. ]] ensure:[blamedProcess terminate] When you run this, MessageTally will tell you that 100% CPU was spent in Delay wait for the blamedProcess. If you look at CPU utilization you should see practically no CPU activity. How come? When the timer event fires, it puts both message tally's profiling process as well as the blamed process in the runnable queue. MessageTally (being at higher priority) will go first and see that the process that is currently active is the blamed process. It will assume that the entire last msec was spent in it since it has no other timing base to use. Be wary that the above can happen in many, many variations and depends heavily on your system's timer accuracy (if you want to play with this, increase the blamedProcess' wait time step by step and compare the reported CPU utilization various platforms). In general, when you see Delay wait show up in your tallies you can be sure there is something wrong with the measure. Unfortunately, there is nothing much you can do about this other than be aware of it and wait. Because we ran into the same issues I have written a better profiler at Qwaq (which is both sub-msecs as well as primitive-aware) and I think we'll be releasing the necessary support for this in Cog but this will still take a while. For comparison, when you run our profiler on the above code you get: Reporting - 45,975 tallies, 5,000 msec. **Tree** 98.9 (4,945) ProcessorScheduler class startUp 98.9 (4,945) ProcessorScheduler class idleProcess 98.9 (4,945) ProcessorScheduler class relinquishProcessorForMicroseconds: **Leaves** 98.9 (4,945) ProcessorScheduler class relinquishProcessorForMicroseconds: So it correctly reports that the code was doing nothing (executing the idle process). 5) Summary ---------- In order to profile your server you definitely need to take of 1) through 3) or else you won't be able to find out anything. With these issues taken care of, just add a remotely executable action that runs message tally for the given number of seconds (we have this set up on our servers so that when you go to /profile?secs=xxx it runs the profiler for that amount of time and spits out the result). Unless you do work that has very fast responses and is designed to work with high load this should be good enough to get you a pretty good understanding of what is going on. Hope this helps! Cheers, - Andreas

2 4

About resourceBaseUr, serverName and serverPath
by Miguel Enrique Cobá Martínez 27 Feb '09

27 Feb '09

Hi all, I have some questions about #resourceBaseUrl, #serverName and #serverPath. In their default state they are ''. This is how I understand them: #resourceBaseUrl ----------------- This is a handy way to *NOT* write your server hostname + resource path every time you render a link or a image so, if #resourceBaseUrl is 'http://www.example.com.mx/resources/" then instead of writing: html image url: 'http://www.example.com.mx/resources/images/image.jpg' you write: html image resourceUrl: 'images/image.jpg' without polluting your code with the server name. This way, you can change the server name to other domain or change the resource directory to another path or even another server. This is used for *you* the developer to generate URLs in anchors and images. Is this correct? # serverName ------------- This is used for *Seaside* to generate URLs in forms, links with callbacks, link and meta tags. If you don't specify it, they are are generated as: http://localhost:8080/seaside/myApp?_s=asdfasdf&_k=adsfasfa if you are accessing the app as: http://localhost:8080/seaside/myApp or http://example.com/seaside/myApp?_s=asdfasdf&_k=adsfasfa if you are accessing the app as http://example.com/seaside/myApp It appears to me that if they aren't specified, Seaside USE the serverName and serverPort from the request. If you specify them, those values ARE used, no matter the name you used to access the app. That is, if you specify this preference, say, as 'error.com', and your app is in 'example.com', the url are generated as: http://error.com/seaside/myApp?_s=asdfasdf&_k=adsfasfa This will crash your app, because error.com doen't exists. Or, another example, if you specify 'example.com' and are accessing the app from your local machine, for instance, when developing, as: http://localhost:8080/seaside/myApp the URLs generated will be: http://example.com/seaside/myApp?_s=asdfasdf&_k=adsfasfa and you will be interacting with your real application and not the development one. Is this correct? # serverPort This one behaves analogous to serverName. So, can you please comment about this preferences and tell me if I am wrong? Thank you, Miguel Cobá

2 1

how to get dynamic files?
by Oleg Korsak 25 Feb '09

25 Feb '09

Hello! My method in a WAFileLibrary subclass returns dynamic contents (composes it on the fly). But when I'm trying to download it (no matter how many times) - the contents is the same as in the begining. Is it possible to get a dynamically created file from WAFileLibrary so it could be different for each request? Thanks.

5 4

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

seaside February 2009