Hi,
Is there a plan to have tODE or at least the GCI library ported to Pharo 5? Cheers, Doru -- www.tudorgirba.com www.feenk.com "Next time you see your life passing by, say 'hi' and get to know her." -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Well, from what I understand, the usage of tODE of Pharo itself is very little so I suspect that part would be easy to migrate if there is anything at all to be done. This is compared to GemTools which involved a lot of changes to OB for example. However, it seems to me migrating to 4.0 would be much easier. Probably because of CGI lib. In Pharo 4.0 we use the same FFI as in 3.0. But in Pharo 5.0 we have these new UFFI. So all the 'GemTools-ClientGCI' would need to be migrated to UFFI as much as I can understand. If we have a couple of tests we could fork tODE (those packages of GemTools are inside the tODE repository) and start giving it a try. Esteban was writing a "migration" document on how to migrate from old FFI to new FFI but I don't know it's status. Dale, do you know who wrote the FFI binding for CGI lib? Is he still around? Cheers, On Tue, Mar 29, 2016 at 12:59 PM, Tudor Girba <[hidden email]> wrote: Hi, You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
In reply to this post by Tudor Girba-2
Yes, there is a plan ...
I have been waiting for Pharo 5.0 to stabilize before jumping in ... also I am assuming that the FFI api has changed significantly -- is that true? If so, the hard work will be porting to the new FFI ... I have recently started work on Tugrik[1] (for Pharo 4.0) and part of the project is to do a cleaner implementation of the GCI libraries, so for Pharo 5.0, I would probably start with the GCI libraries from Tugrik, port them to Pharo5.0 and then port tODE to using the new GCI libraries ... With all of the other work I am doing, I don't know when I will get around to Pharo5.0, but if you are interested in helping, I would be willing to split the GCI library work out from Tugrik into a separate project and help with the sticky bits in the port of the GCI libraries to Pharo5.0 ... then tODE would follow on once the GCI libraries were ported ... Dale [1] https://github.com/dalehenrich/Tugrik On 03/29/2016 08:59 AM, Tudor Girba wrote: > Hi, > > Is there a plan to have tODE or at least the GCI library ported to Pharo 5? > > Cheers, > Doru > > > -- > www.tudorgirba.com > www.feenk.com > > "Next time you see your life passing by, say 'hi' and get to know her." > > > > -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
On Tue, Mar 29, 2016 at 3:58 PM, Dale Henrichs <[hidden email]> wrote: Yes, there is a plan ... wow... is Tugrik the idea you talked to me years ago with Martin? (In Edinburg) With all of the other work I am doing, I don't know when I will get around to Pharo5.0, but if you are interested in helping, I would be willing to split the GCI library work out from Tugrik into a separate project and help with the sticky bits in the port of the GCI libraries to Pharo5.0 ... then tODE would follow on once the GCI libraries were ported ... Dale, I personally think that having the CGI libs in a separate repository makes A LOT OF SENSE, wether we will gonna give it a try right now or not. See...GemTools needed them. Then you copy them and put them inside tODE. Now you need it too for Tugrik. And now we may want to fork it and make it work in Pharo 5.0 ;) As part of my work with the Pharo Consortium and OSSubprocess I did work with the new FFI. So I can give a hand here. You even have tests for it: https://github.com/dalehenrich/Tugrik/tree/master/repository/GemStone-GCI-Tests.package/GsGciTest.class/instance So... I think it's a matter of defining a new home for CGI lib, fork it, and start giving it a try in Pharo 5.0 with new FFI. If you move these guys into a separate repository (on github) that I can load them into a fresh Pharo 5.0, I will do my best to do the first try when time allows me. Note that that doesn't mean you have to remove them from Tugrik, updates it ConfigurationOf etc...we can do that later. Just copy them to a new repo and once we have tests working in Pharo 3/4 then we start with 5.0. Cheers, Dale You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
On 03/29/2016 12:15 PM, Mariano
Martinez Peck wrote:
I think that Tugrik is probably a slightly different take on the idea that was tossed around in Edinburg, but I think that it is a more practical one --- namely, I'm pretty sure that I can duplicate the functionality of MongoTalk with TugrikTalk at the Smalltalk API level, while implementing the persistence and query functionality in GemStone.... which I think can be very interesting :) Oh, I agree --- the GCI used by GemTools and tODE date from 2007, and I think they deserved to be rewritten before being shared and the Tugrik project became the opportunity to rewrite the GCI implementation ... Yes, this is a critical component ... I also think that with SmalltalkCI, I will (eventually) be able to run these tests on travis:) I didn't know that you did work on the FFI, expert advice would really jump start the project ...
I'll get started on the split right away ... getting smalltalkCI to run the GCI tests on travis will be a bit of a challenge --- another of the issues that is on my plate --- but it is a solvable problem and if you are taking a crack at porting the GCI I can work on getting the infrastructure in place as I am planning to work a bit more on SmalltalkCI right now anyway ... I'll send mail when the split is complete ... Dale -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
In reply to this post by Mariano Martinez Peck
Mariano,
Barebones split ... no baseline, but you can look at the Tugrik baseline for load order ... the GemStone-GCI-Server-Tests.package should be loaded into GemStone as it provides a set of methods that are called by the client-side tests ... I think this should be enough to get you going --- when you have a chance ... Dale [1] https://github.com/GsDevKit/GemStone-GCI On 03/29/2016 12:15 PM, Mariano
Martinez Peck wrote:
-- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Hi Dale, Ok, I started with this. I created a baseline, I refactored a bit the session description used for unit testing and added some help to the README (this is temp help notes but I wanted to write them down now while I am at it than later...). Following the instructions you can find in the README, I managed to run some tests, but only #testGciLogin and #testGetGciErrSType passed. All the rest of the tests are still red. I am still talking on Pharo 3.0, obviously hahahha. Any idea why would the rest fail? Do they work for you in your Tugrik setup? On Tue, Mar 29, 2016 at 5:13 PM, Dale Henrichs <[hidden email]> wrote:
You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Do you have some error stacks? AFAIK they all passed .... it has
been awhile since I ran the gci tests explicitly as I was working on
the server side support ... but running the Tugrik tests excercises
most of the GCI ... a stack will job my memory ...
Dale On 03/29/2016 02:35 PM, Mariano
Martinez Peck wrote:
-- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
I just ran them in my Tugrik image and they all passed ...
On 03/29/2016 02:57 PM, Dale Henrichs
wrote:
Do you have some error stacks? AFAIK they all passed .... it has been awhile since I ran the gci tests explicitly as I was working on the server side support ... but running the Tugrik tests excercises most of the GCI ... a stack will job my memory ... -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
In reply to this post by Dale Henrichs-3
Ahhhhh, now I remember ... I had to hack on the FFI code to get my
subclasses of ExternalStructure to initialize correctly ...
I'll isolate those changes and pass them along On 03/29/2016 02:57 PM, Dale Henrichs
wrote:
Do you have some error stacks? AFAIK they all passed .... it has been awhile since I ran the gci tests explicitly as I was working on the server side support ... but running the Tugrik tests excercises most of the GCI ... a stack will job my memory ... -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
In reply to this post by Dale Henrichs-3
Nope, the FFI hacking was not it ... I had done some class renaming
and the FFI references to those classes were renamed incorrectly and
I have uncommitted changes in my Tugrik image ... I'll push those
changes out shortly ...
On 03/29/2016 02:57 PM, Dale Henrichs
wrote:
Do you have some error stacks? AFAIK they all passed .... it has been awhile since I ran the gci tests explicitly as I was working on the server side support ... but running the Tugrik tests excercises most of the GCI ... a stack will job my memory ... -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
In reply to this post by Dale Henrichs-3
done ... btw the sigabort tests will fail (after a hang) if you have
any gems logged into the test stone ...
Dale On 03/29/2016 02:57 PM, Dale Henrichs
wrote:
Do you have some error stacks? AFAIK they all passed .... it has been awhile since I ran the gci tests explicitly as I was working on the server side support ... but running the Tugrik tests excercises most of the GCI ... a stack will job my memory ... -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Thanks Dale. Ok, with your commit, the only failing test is #testSigAbortHandlingNew. And then, both tests of GsGciTsTest raise "Cannot return the given type" to me. Does this also happen to you? Cheers, On Tue, Mar 29, 2016 at 7:10 PM, Dale Henrichs <[hidden email]> wrote:
You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Does the new FFI block the image on function calls? If so, then the
thread safe libraries aren't that useful, since they don't currently
include non-blocking variants. The point of the thread safe library
is allow one to safely make concurrent GCI calls, but if a single
FFI calls blocks the image, you can't have more than one in progress
anyway ... VW arranges to make FFI calls in a separate os thread, so
blocking the thread is not an issue ...
We can leave the thread safe stuff out of CORE group and not bother with it ... Dale On 03/29/2016 04:12 PM, Mariano
Martinez Peck wrote:
-- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
On Tue, Mar 29, 2016 at 8:27 PM, Dale Henrichs <[hidden email]> wrote:
No, threaded callouts are not yet supported.
Yes, that's a good idea.
You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Cool!
I am working on trying to figure out a way to make smalltalkCI run gemstone and pharo, so that travis/smalltalkCI can be used to run the Pharo-based GCI tests ... Dale On 03/29/2016 04:32 PM, Mariano
Martinez Peck wrote:
-- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
On Tue, Mar 29, 2016 at 8:36 PM, Dale Henrichs <[hidden email]> wrote:
Yes, that would be nice!!! BTW, I just pushed the missing bit in the readme of copying the libcgi* libs. Those steps would be the requiered ones for Travis. BTW2... it would be nice if: 1) You can also build the custom todeClient so that you know from where to copy the libcgi. Or even better... call your download script from gsDevKit_home to put the cgilibs in the pharo 3.0 vm. 2) You can replace my instructions on how to load cgi into gemstone with a todeIt script :) I know how do to it for Pharo ;) That way we can also automate all that. BTW3 In my original tode vm image I have a very large hierarchy of GciLibrary (GciLibrary allSubclasses size -> 7) for all combinations of of Gemstone 3.2, 2.2. etc etc. But with the project we are working now, we ONLY have GsGci32xLibrary. So....that means that this new CGI wrapper will only work for 3.2..... do we agree??
You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
On 03/29/2016 05:19 PM, Mariano
Martinez Peck wrote:
Yes --- GsDevKit_home has a script for adding the required libraries to the known Pharo vm directories, but the challenge at the moment is to be able to arrange for the smalltalkCI pharo/run.sh code to be used for running the pharo test nested in a gemstone/run.sh .... Yeah the individual pieces are straightforward to automate, but doing the automation within the smalltalkCI environment is challenging ... I don't want to have to duplicate the pharo/run.sh code inside the gemstone/run.sh script ... The new GCI code is intended to support GemStone 3.2.x and later ... the Tugrik code is running in 3.3.0 ... but then that is why I want to get the travis tests working so we can verify against 3.2.x and 3.3.x ... Dale -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
OK, I will let you see if you can figure it out the travis stuff. I will be fighting against new FFI. For the record, I opened a new branch for that: https://github.com/marianopeck/GemStone-GCI/tree/50 Of course, not a single test is running yet ;) On Tue, Mar 29, 2016 at 10:17 PM, Dale Henrichs <[hidden email]> wrote:
You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
On 3/29/16 6:31 PM, Mariano Martinez Peck wrote: > OK, I will let you see if you can figure it out the travis stuff. On my drive home, I figured that I will need to be able to create Tugrik clients that have similar needs to tODE clients (gci libraries and session descriptions), but will not have tODE installed ... so I will probably end up adding support for creating generic pharo clients (along the lines of`createClient -t pharo Pharo-50 tugrik`) and then I will have to add support to the gemstone smalltalkCI for running tests against these custom clients using smalltalkCI ... Dale -- You received this message because you are subscribed to the Google Groups "tODE" group. To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email]. For more options, visit https://groups.google.com/d/optout. |
Free forum by Nabble | Edit this page |