Dave wonders: "Ingo Rammer spotted another .NET developer who started a weblog. Something's going on here."

For clarification, yesterday's .NET bloggers aren't from Microsoft. They are just somewhat "well known" in the .NET community due to their great number of valueable contributions [well, ok ... and some occasional rants and raves ;-)] in some mailing lists or newsgroups. That is, we are just the users ;-)

And it's going on ... yet another .NETter who got a weblog. Welcome Tomas!

Ok. Now I read a little bit more of Web Services/SOAP and CORBA. It seems that the author tries to compare them for classic distributed applications (like customer management applications in a corporate LAN). In this case he is probably right that SOAP isn't the best choice - but this depends on the degree of interoperability with other programming languages and platforms which may be needed for an application.

However, I'd really like to know how many cross-company applications which interact by using CORBA have been deployed in the last years ;-). How many based on SOAP or XML-RPC?

Web Services/SOAP and CORBA: This paper is written from a CORBA point of view and the author tries to trash Web Services. I have to admit that I did not yet read the whole thing but I found one interesting bit: In section nine, the author talks about the possible advantages of human readable data formats

Some may argue that it is a good thing that the protocol is readable by humans. This may allow for easier developing and debugging, but is that true? What real use is human readability at the protocol level? In any inter-corporation data exchange scenario it’s going to be encrypted anyway. And are developers really interested in what passes over the wire? I’d rather not think about that and concentrate on the important parts of the application.

Maybe we should think again about SMTP and POP3 and how they changed the world, right? Human readable protocols are normally a little bit slower but somehow people actually like to see what's going over the wire. They maybe also like to be able to debug email systems simply by using telnet. And finally: if I can read it, I can implement it in no time.

Did you ever write your own CORBA implementation for a programming language of choice, where no implementation existed before? Why not?

Sam Gentile seems to like Radio: Wow! 5 minutes to up and running! Ingo Rammer, you are right. This is much easier!

Another .NETter is about to start a WebLog. Welcome Sam! And by the way, you really should look into using Radio, it's just way more comfortable.

Ok. I figured out how to persuade Radio Userland to render the latest day's posts in a form that can be included directly on the main page of http://www.dotnetremoting.cc. If anyone's interested how this works, just tell me and I'll write a short summary about it.

Gartner.com, 3/25, Note FT-15-9819: Gartner Warns Oracle Customers to Check License Fees. It seems as if some Oracle salespeople charged their customers a little too much. Just think about what would have happened if Microsoft did things like this?

[CNET] Hasso Plattner, co-chairman of SAP says "This industry will not survive in the next 20 years with applications built by a few big companies," Plattner said during his keynote speech. "This industry will only survive if there is a community of developers building applications that can work together across many systems."

Ok. I agree that interop is a Good Thing. Somehow Plattner sounds quite like Dave when he tells the world, that SmallCo's will have to stand up and work together. Funny thing that Platter comes from one of the larger BigCos ...

I nevertheless believe that Plattner doesn't really care about interop. I guess his only challenge is to keep SAP sales in a market that's somewhat saturated - but ok, as long as his way to reaching the target is the same as ours, it's fine by me.

Rotor is brilliant!

There are large parts of the .NET Remoting framework inside. The most interesting are in sscli\\clr\\src\\bcl\\system\\runtime\\remoting and in sscli\\managedlibraries\\remoting\\channels and remoting\\metadata. Wow! There is just no comparison between getting your information from the ECMA spec versus reading these sources instead.

If you're interested in the inner workings of this framework, you just have to download Rotor - it's only about 10 Megs and it's way easier to read than an Anakrino decompiled version [albeit you really get used to it and it's still a great tool!]

Oh yes, and by the way, you'll see things of which Anakrino doesn't know about. The handling of remote objects inside the CLR f.e., how __TransparentProxy is used, etc. This is just brilliant.

.NET Remoting: Today Peter posted his Jabber channel for Remoting. It allows you to use the jabber instant messaging infrastructure for distributed applications. Cool stuff, indeed.

[...] special thanks are due to Ingo Rammer for helping us out with some gnarly Remoting plumbing questions [...]. You're very welcome!

btw. I posted this about eight hours ago but somehow Radio didn't upstream it ... after restarting Radio right now, this and another post have simply vanished without traces. Mysterious.

The full source for a CLI implementation which is strangely similar to the CLR can now be downloaded, perused and extended [Peter Drayton's Radio Weblog]

Ok. I guess today's evening program has been defined with this release ... ;)

I need an easy to use "todo" list. Maybe I just need to make a "todo" category. [deem] In fact that's exactly what I do. All the aggregator posts for which I don't have the time to read them (at the instant I see them) get posted to a "must read" category that's not upstreamed ...

Annother Microsoft blogger: Wolfgang Manousek. This time it is an actual developer. [deem]

Welcome! Just be sure not to violate corporate policy by telling us too much about your heap corruption bugs. Well, in fact do tell us ... at least I like to hear what's going on inside.

By the way, Mike do you already get sales commision from Dave? ;-)

Some days ago, Sami Jaber of http://www.dotnetguru.org asked me if he might translate my article on the performance of .NET Remoting vs. ASP.NET Web Service to French - of course I agreed.

So if you're French, you can now read this translated version here. [yes, I really understood parts of it as I learned French for about four years ... ;-)]

Un grand merci au passage à l'auteur de l'article original : Ingo Rammer, spécialiste de la plateforme .NET et auteur de l'ouvrage "Advanced .NET Remoting". You're very welcome!

The DevCon was great, I had a blast, .... [Simon Fell]

Somehow too bad that it's been too far away from here ... (if you wonder, I'm in Austria, Central Europe)

Cool!  Dave now published a Text-editing cheat sheet.  Thanks [Sam Ruby's Radio Weblog]

Thanks from here as well!

Wow! That's customer service. It's sunday and Jake from Userland fixed my bug in fewer than 24 hours. Let's face it, this software is $39 - and I somehow didn't expect this bug to be fixed that quickly. Incredible. Thanks!

Dave, you can really be proud of your team!

While playing with Radio during the last days, I noticed what a lot of other radio users must have noticed before: There is no configuration setting for the date format but instead the date/time format of the local machine is used when rendering the pages. This might not be the best idea if someone like me, who's using a German W2K box, is writing an English-language weblog, right?

Well, I wrote a little piece that shows how to manually format the date for each day.

Dave already has "[...] an idea what change caused this problem." - That's been fast.

I'm still playing around with Radio - and some problems really bug me right now. This is also the reason why the calender on the right side doesn't work yet.

Oh, and yes all the old links still work and will continue to do so!

I'm on the train again. This time I'm travelling back from Linz, Austria where I did a talk on .NET Remoting at the computer science department on invitation by Dr. Moessenboeck [1]. There were about 30 to 40 people, mostly assistent professors and students and appearently most of them already had been working with .NET before which provided a good basis for this talk.

The talk went quite good and I guess it has been quite positively received. This is for example unlike to Technical University of Vienna where Linux rulez. In Vienna I've been to a talk by Rafal Lukaviecki and he got attacked [verbally] personally by some open source freaks just for talking somewhat positively about Microsoft.

In fact, I don't take it religiously and I like it if my audience does the same. I really like .NET but I regularily use Java, and yes, even Linux. In fact the largest systems I have designed and developed currently run a backend on Open Source software (Tomcat 4) on a Solaris machine. The front ends nevertheless run on Windows - XML made it easy. So I guess religious defenders of certain operating systems or programming language might just miss some opportunities ...

After this talk Mr. Moessenboeck and his .NET team invited me for dinner. We chatted for some hours and they reminded me again how much fun research work can be. The really do some cool stuff there - too bad that I didn't finish my studies which somehow prevents me from joining such projects. Every time I talk to people from the academic community, my idea about starting studies again gets a little bigger.

Too bad that you cannot just omit the CS 101-style courses and start right with the interesting stuff in years three or four ... ;)

[1] Hint: If you ever worked with the Oberon system or language, you should know this name ...

Today I had a somewhat interesting discussion with one of my co-workers whom I really respect. He has been working in the IT business as a software developer and project manager for ages now [1] - in fact he's one of the guys who used punch cards in "real applications" and later tended to walk to their customers with a stack of several hundred pages of printed mainframe assembler ...

Today we discussed about the complexity of the current mainstream programming languages when used for the development of business applications. He's been working for some years now with 4th generation languages like DataFlex (starting back in the dark ages of DOS & co) and he said that he "just didn't have to care about data types, databases, and so on - it just worked". Today - when using DataFlex - he said that he can even develop web applications in a snap.

Don't get me wrong here - this guy is absolutely ok and very interested in Java and .NET - it's just that he says that with both of these choices, the development effort and even more the quality assurance and maintainance effort for business applications actually increased.

In fact, shouldn't have software development become somewhat easier and faster in the last twenty years? 

Maybe we're all just wrong ;-)

[1] Any resemblance to Sid from userfriendly in just purely coincidental ...

John "was away working on yet another secret project [...] life-changing [...]". I can only confirm. He told me a little about it and this one definitely will turn out quite interesting [that is, for the ones working there]. It's somehow like working for a startup with great prospects again. But I guess this one isn't a dotbomb ... 

He also noticed that I've been playing with "ICorProfilerCallback and friends [...] I have some sample code kicking around that does just that, so I'll post it once I've had a chance to clean it up a bit." Great! Hurry up! ;)

When questioning Emacs as a blogging tool some days ago, I knew that one of you would respond. I just didn't think that it would be Peter: "I might even be able to switch completely back into Emacs and declare this weblog a "Radio Free Radio-free Zone"? <g>" 

Ok, ok ... I sometimes tend to use Emacs as well but for me it's yet another editor - nothing religious about how one writes his code.

But after all it's still better than vi. [I bet there's someone blogging with vi who'll send me an email in about ten seconds ;)]

By the way, I just received some little gifts from Redmond. The best one is this signed T-Shirt from the development leads for Remoting and Security (Brian and Peter) and some of their developers: 

This is just too cool. Thanks guys!

I currently have absolutely weird e-mail problems (for about two weeks now). Some outgoing mails just disappear without notice and some incoming mails get bounced to their senders just to work when resent some minutes later. If you sent me an email during the last days to which you expected a response, could you please resend it? I generally answer each and every email I get [apart from spam ;)]. Sorry for this ....

btw. someone sent me the error message he got:

- BREAKING NEWS: Amazon has published the final price of my book. It's 30% off now. Click here to pre-order.

- 11:55 pm CET. While ego-surfing on Google, I noticed that some of my source code has found its way into IBM Websphere Portal Server. Does that mean that I may say that "even IBM's using sourcecode I've written" ;-)?

- 08:55 am CET. Peter tries to convince Jeff to keep blogging after his book is finished: "just download Radio, CityDesk, Blogger, Blosxom, Emacs, basically anything, and keep on blogging". 

Agreed. But does anyone here use Emacs? For blogging? For real?

- Peter also shows a nice point for why JScript.NET is the new VB.

- 08:33 am CET. I just received an email from Serdar in response to yesterday's "PDF with index server?" question. He pointed me to a filter at http://www.adobe.com/support/downloads/detail.jsp?ftpID=1276. Thanks!

- 9:55 pm CET. Huh. I guess I should really look into Jabber. Today two people from the blog community contacted me independently and told me that they're working on a Jabber channel for .NET Remoting ;-)

We're in blogworld, you guys should really talk to each other and save yourselves some work ...

- UPDATE: They're working on the same thing. They just contacted me independently ;-)

- Simon's been faster. Today I read Peter's post about "just [adding a] layer on the searching & caching value" instead of converting a filesystem to a database. I really wanted to point him towards Index Server but I didn't have blog access in the office so Simon got the point first ...

By the way, I recently heard that index server doesn't support PDFs with full text indices. Is this true? Does anyone know for sure? (just blog or drop me a note at rammer@sycom.at).

- 10:14 pm CET. As Peter mentioned yesterday, working ICorProfilerInfo::SetILFunctionBody is "Cool!". But it comes at a prize. What I did some days ago was to replace the IL body with a predetermined destination IL.

Of course this is not what I'm finally trying to achieve. Instead I want to inject a call to a foreign method into precompiled binaries. This also means that I have to add an assembly reference to each assembly that's about to be dynamically modified.

I'm currently about to search for any hints on how to get the IMetaDataAssemblyImport and the IMetaDataAssemblyEmit interfaces from a running assembly from within ICorProfilerCallback. This is mostly a trial-and-error game as the documentation isn't really that helpful and I'm still unsure if I wouldn't have to use IMetaDataEmit instead and therefore append a module ref to a module instead of working with the assembly itself.

If you are from Microsoft: do you know if there's any current documentation of these interfaces (and their relation to ICorProfilerCallback) available that's not in the Framework SDK or Tool Developer's Guide? ;-) (also, if you do know something about this which might be helpful, could you drop me an email at rammer@sycom.at?) [ok, just kidding ...]

- 00:35 CET. Argh! Documentation really is bad here. After browsing through it for another hour I finally went back to the headers and found the needed method IMetaDataDispenserEx::FindAssembly() ...

- 01:06 CET. Even more argh right now ... still wondering how John could figure this out ;-). [ok, maybe he has the slight advantage of knowing C++ ... ]

- 01:18 CET. Offline for today. Enough is enough ;-)

Wow! In the last ten programming hours I learnt more about the C++ than ever before.

As I said yesterday, I'm currently playing around with MSIL-injection during the JITter phase. Unfortunately this is only possible with C++ which I've never ever used before ... seems like a bad point to start. In MSDN Magazine's December 2001 "Under the hood", Matt Pietrek said the following about this topic: "See [...] for details, and don't expect it to be easy!". In fact, even the Framework SDK's Tool Developer's Guide says "Using ICorProfilerInfo::GetILFunctionBody and ICorProfilerInfo::SetILFunctionBody is not trivial.". Well, I confirm both.

In addition, I want to say "Using C++ is not trivial" ... but you already knew this ;-)

During the last 10 hours I've been writing some really weird source code. If someone would have shown it to me a week ago, I wouldn't have understood even the tiniest bit. I simply hope that I can still understand it in one or two weeks ...

Ok ... now to the real point of this post. If you ever want to play with this functionality, mind the following points:

• If you want to use std::cout, you have to link with libcpmtd instead of libcpd so you will have to change the makefile if you're starting from the sample profilers (ok, you already knew this but it's been news for me ...)
• The Tool Developer's Guide has some errors. Always go for the .h files instead
• If you're new to C++, you better have a friend like Christian Wehrl (Gumbo) who can give you a "pointers in C++"-crash course over the phone within 15 minutes. Thanks for this!
• You better also have a girlfriend like Katja who hands you her C++ book. Thanks as well!

The only problem is that I [don't laugh] have never written any C++ before, so I'm currently iterating my way in a trial-and-error fashion through adopting the GCP-Profiler sample from the Framework SDK. I've got some help in form of the brilliant Accelerated C++ book which has been recommended to me by people from [DOTNET]. And yes, I also have Stroutrup's bible here ...

The main problem is that this thing isn't plain C++ but instead it's something along the lines of multithreaded COM DLL. This really got the fun started for me as a C++ non-programmer.

I guess I can now imagine how COBOL programmers must have felt, when they were told to switch to developing Windows GUI applications after years of mainframe hacking ... this simply is a paradigm clash.

I want my highlevel languages back ... compared to this, ILAsm really is on a very high level.

Anyway, I'll keep you updated. [Please stop the giggles now, ok? ;-)]

- 01:24 am, CET: I finally got the IMAGE_COR_ILMETHOD. The bytecode is only some bytes away now. I still don't really like C++, too many &s and *s ;-)

I finally managed to bring the complete introduction to my upcoming book Advanced .NET Remoting online. You can find it here. The book got a little longer than originally planned and is now a little more than 400 pages (I haven't seen the final page count, including TOC and index yet, but the  real content alone is 392 pages). And by the way [for some advertisment], it's now available for pre-order at amazon.com.

I also did a redesign of this site - hopefully it's a little bit easier to navigate now ;-)