Re: [Zope-dev] Help review #181754
On 2010-7-21 07:32, Wolfgang Schnerring wrote: * Wichert Akkermanwich...@wiggy.net [2010-07-20 19:28]: On 2010-7-20 18:15, Christian Theune wrote: At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application. I'm not sure I agree. Like everything else servers have bugs, so it can't hurt to test how your application would behave given certain server bugs. I don't think it is usually a productive assumption that lower layers fail to uphold their end of the contract. Maybe an extrapolation/hyperbole illustrates my opinion: Cosmic rays might also flip bits in your computer's RAM or disk, but I don't think it's worthwile to test how your application reacts when the python interpreter (or whoever, really) presents it with mangled data structures or objects or whatnot. And for some situations you do want to explicitly test for such things. It all depends on how critical your app is. I'm not sure I'm not the only one who has tested code with randomly broken/bit-flipped input to test robustness. Wichert. -- Wichert Akkerman wich...@wiggy.net It is simple to make things. http://www.wiggy.net/ It is hard to make things simple. ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Help review #181754
On 07/21/2010 09:17 AM, Wichert Akkerman wrote: On 2010-7-21 07:32, Wolfgang Schnerring wrote: * Wichert Akkermanwich...@wiggy.net [2010-07-20 19:28]: On 2010-7-20 18:15, Christian Theune wrote: At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application. I'm not sure I agree. Like everything else servers have bugs, so it can't hurt to test how your application would behave given certain server bugs. I don't think it is usually a productive assumption that lower layers fail to uphold their end of the contract. Maybe an extrapolation/hyperbole illustrates my opinion: Cosmic rays might also flip bits in your computer's RAM or disk, but I don't think it's worthwile to test how your application reacts when the python interpreter (or whoever, really) presents it with mangled data structures or objects or whatnot. And for some situations you do want to explicitly test for such things. It all depends on how critical your app is. I'm not sure I'm not the only one who has tested code with randomly broken/bit-flipped input to test robustness. That double negative confuses the hell out of me. :) However, I don't think the request of the given bug matches that requirement anyway. Christian -- Christian Theune · c...@gocept.com gocept gmbh co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
[Zope-dev] Help review #181754
Hi, I just rejected issue https://bugs.edge.launchpad.net/zope.app.testing/+bug/181754 There might be room for different oppinions and if anyone would like to veto the decision, I'd welcome a discussion. Christian -- Christian Theune · c...@gocept.com gocept gmbh co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Help review #181754
On 2010-7-20 17:39, Christian Theune wrote: Hi, I just rejected issue https://bugs.edge.launchpad.net/zope.app.testing/+bug/181754 There might be room for different oppinions and if anyone would like to veto the decision, I'd welcome a discussion. Your rejection contains an assumption that the request won't make it to the application. Is that assumption testable? Wichert. -- Wichert Akkerman wich...@wiggy.net It is simple to make things. http://www.wiggy.net/ It is hard to make things simple. ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Help review #181754
On 07/20/2010 06:01 PM, Wichert Akkerman wrote: On 2010-7-20 17:39, Christian Theune wrote: Hi, I just rejected issue https://bugs.edge.launchpad.net/zope.app.testing/+bug/181754 There might be room for different oppinions and if anyone would like to veto the decision, I'd welcome a discussion. Your rejection contains an assumption that the request won't make it to the application. Is that assumption testable? I guess so. I think it's relatively clear that the application server should insulate the application from malignant requests. The details probably depend on the HTTP implementation. zope.server, zope.app.twisted, Paster and others might deal with this differently. Looking at the interfaces between those parts of the application it makes no sense to have syntactially invalid requests end up on the application level. At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application. Christian -- Christian Theune · c...@gocept.com gocept gmbh co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Help review #181754
On 2010-7-20 18:15, Christian Theune wrote: At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application. I'm not sure I agree. Like everything else servers have bugs, so it can't hurt to test how your application would behave given certain server bugs. Wichert. -- Wichert Akkerman wich...@wiggy.net It is simple to make things. http://www.wiggy.net/ It is hard to make things simple. ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Help review #181754
On 07/20/2010 07:28 PM, Wichert Akkerman wrote: On 2010-7-20 18:15, Christian Theune wrote: At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application. I'm not sure I agree. Like everything else servers have bugs, so it can't hurt to test how your application would behave given certain server bugs. Right. IMHO this specific bug asked for would usually mean that the server breaks and can pass anything to the application. Whatever happens: the server needs to prepare a data structure to hand over to the application. If it can't do that, then there's nothing to test what happens with the application. I guess this kind of test (what happens with the application if it receives invalid data) needs to work around server/publisher components and inject data directly to application methods. Christian -- Christian Theune · c...@gocept.com gocept gmbh co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Help review #181754
* Wichert Akkerman wich...@wiggy.net [2010-07-20 19:28]: On 2010-7-20 18:15, Christian Theune wrote: At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application. I'm not sure I agree. Like everything else servers have bugs, so it can't hurt to test how your application would behave given certain server bugs. I don't think it is usually a productive assumption that lower layers fail to uphold their end of the contract. Maybe an extrapolation/hyperbole illustrates my opinion: Cosmic rays might also flip bits in your computer's RAM or disk, but I don't think it's worthwile to test how your application reacts when the python interpreter (or whoever, really) presents it with mangled data structures or objects or whatnot. To put it differently, yes, servers have bugs, but the place to fix them is in the server. If we can't rely on our layers/abstractions to hold, I feel we lose most if not all benefits of having them in the first place. Wolfgang ___ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )
