There's a fix for the context which involved an mcp search and execute gateway. Essentially the mcp server queries for desired capabilities, gets search results with execution and requirement details and fires off the actual mcp as subcalls:
You can call it "rag for mcp". I was pushing it hard a few months ago and nobody seemed to care but I'm all in if the timing has caught up to the tech.
It's nontrivial effort: basically a giant survey of all the mcp servers, running inference over them to figure out how to instrument them, cross referencing to make sure they are the "official" sources (or at least the ones that search engines think are) then using qdrant to do embeddings and reranking and offering it for free.
If people have become interested I'm all in. I'll bring the infra back up. I just don't want to spin my wheels on dead end streets.
The value proposition is solid, the problem is real, this fix works, it's fast, it's free, and people give exactly zero shits. I dunno...
"VC++6 is remarkably powerful for 1996. It has features such as "Go to definition", breakpoints, stacktrace, and variable inspections (but no Intellisense auto-completion yet). I never used it but it must have felt like a dream at the time."
And here we are, in a generation of people writing blogs that never used VS6. I am now officially old.
I was still using VS6 as late as 2009 btw...also it's from 1998. If you made a list of Microsoft bangers it's in the top 5 with probably windbg, quickbasic and windows 3.11.
Which is why, I usually assert I cannot understand the nostalgia of CLI and TUI, being there at the time, and not being able to use some of these systems, due to the amount of money they required.
The appeal of the cli/tui is because agents can use them better and because people want to ssh in and just use something without any fuss or lag
And I know the hn reply, it's "what do you mean? all you have to do is this other thing over here" and that's the point ... people don't want to. So they don't.
> Which is why, I usually assert I cannot understand the nostalgia of CLI and TUI, being there at the time, and not being able to use some of these systems, due to the amount of money they required.
I was not there at the time, but one appeal of CLI (not TUI) is scripting. After a while, you have all your routine packaged in nice alias and commands. And that’s universal across all languages and projects.
Turbo Pascal had breakpoints, variable inspections in the late 80s. I think it had stack traces too but not 100% sure.
I am not old enough to have used it professionally, but my teacher used it for teaching intro programming in the early 2000s. So I used it quite a lot, the debugger was great and the development loop was so tight. Not until I got into web dev did it ever feel "fast" to make change->see change. To this day it is still bad in most stacks.
The earlier Microsoft compilers included since 1985 the debugger CodeView, which could do all that and much more.
Around 1990, the development tools offered by Borland and Microsoft for C and C++ were pretty much equivalent and they both were quite good.
While the Borland languages were like "Turbo-X", the Microsoft languages were like "Quick-X".
The greatest difference between the commercial software available at that time and what exists today is that everything was accompanied by a set of high quality manuals that could teach you anything that one would want to know. Nowadays the quality of technical documentation is usually much worse.
That wasn't the banger for vs6 it was the workflow and muscle memory of the thing. The flow is still unmatched IMHO. It was like avid or photoshop for writing windows software.
Default keys in modern IDEs are basically still the vs assignments from the 5/6 era.
It was the closest Microsoft ever came to making their own emacs or vim. vs6 was like 90% of my screen time as a windows dev in the 90s and 2000s
I've been a linux user for 30 years ... I never had the vs6 level of efficiency in linux, still don't. NetBeans was the closest ... yes, NetBeans... (I've given up though, I do things in nvim, tmux and suffer)
Turbo Pascal was amazing for its time. As a young person learning programming it was a step change in functionality. Before that on PCs you were using Basic or assembly It was cheap and incredibly useful.
Yes, but if you compare the complexity of (Turbo) Pascal to the complexity of C++... language, environment, libraries and cross-compilation...
(A nice thought-experiment is to ask if Quake could have been coded in TP at all - even if memory hadn't been an issue (I think there was no DOS extender for TP, but I could be wrong).)
In the storm of Doom-Quake mania of the mid 90s there was Chasm: The Rift by a small Ukrainian company Action Forms. And if memory serves me right, it was created in Turbo Pascal. It was late in development and came out in 1997 after Quake, so it didn't get much traction. But the engine, though pretty limited, could produce 3D enemies with interesting effects not found even in Quake.
So Turbo Pascal (with a whole bunch of x86 asm inclusions) was totally capable of producing Quake-level games. I myself, in the late 90s, discovered the hidden capacities when I learned x86 assembly from Peter Abel's book. Once I got rid of the primitive TP BGI library and switched to VGA 13h, it was an unbelievable level up in abilities to manipulate pixels on the screen!
I might be misremembering but I thought it was more of a Doom-style engine with 3d models instead of sprites for the entities, rather than a full 3d engine like Quake.
It was definitely something unorthodox - I remember being confused about how it actually worked.
The first time I played, I thought "It is just like Quake". Then you start to notice that the levels are pretty limited - it is all narrow corridors with rare small chambers and open spaces, no steep elevation changes, no rooms above rooms... Kinda like Doom. But then, father's examination shows sloped surfaces, shelves and bridges - stuff the Doom engine can't do.
Maybe, for level geometry, it is closer to portal rendering engines like Build. Still looks pretty claustrophobic even compared to Doom and Duke Nukem 3D. I feel the Chasm level designers could've got more variety from that technology, but again, maybe that was some fundamental limitation. Or maybe it is just like FPS were designed in the mid 90s.
Monsters, items and weapons are fully 3D, though. With dynamic lighting!
Rational Apex Ada is another dev platform that was way ahead of its time in early to late 90's. Multi-user hosted dev environment with incremental compilation and dependency tracking, syntax and semantic error highlighting, semantic search (i.e function signature) across whole repo, its own version control system with a git submodules style structure, automatic formatting as you write code. [Remote] Debugging and emulation features (stack trace, line of code, disassembly, etc), plus excellent VxWorks integration and tooling. Not to mention all the Ada language features which are still not available in modern languages.
I used VS6 professionally and for private business around 2000-2004, and it was still going strong then. VC++ was great.
One thing though that I still have nightmares about is Visual SourceSafe, Microsoft's idea of a source control system for small teams. It was not only terrible to use (and slow), but we regularly lost data in it due to concurrency issues.
Speaking of MS and source control, I have to shout out this incredibly niche channel [1] that recently covered "Microsoft Delta", a precursor and home grown effort that was eventually abandoned in favour of buying in what would become SourceSafe.
It was my favorite VCS ever at my first workplace where we deployed .war to prod tomcat from eclipse with one click. No tests, no PRs, no tickets. Customer would call me and I could get a change out to them within 5 minutes. Most (and only!) agile workplace I ever experienced in two decades.
The load times of VC6 versus VSCode... yeah. I would bet you can load a VC6 project faster in OP's environment faster than VSCode could load the same project in a current environment.
The last time I used it in anger to release commercial software was round about the year 2020, at which point the dev environment for that particular piece of software that customers were still paying annual license fees on was a VM machine. The source code repo it linked to had been unknowingly destroyed years earlier, so the VM image was copied around as needed. One had to find the very latest version of that image, because otherwise any changes one made would of course exclude some other recent changes and customers would receive a Frankenversion.
Starting the VM would reveal a desktop with VC++6 already open, and enough supporting evidence to show how to build the software. Make your changes, build, carefully extract the binary to send to the users, freeze the VM again.
I expect it's still there, still being brought back every year for "one last update."
It feels now like an alternative timeline, one which performance optimisations were first and foremost still. Sometimes I fantasize, thinking how would our current development ecosystem look like, if we never abandoned the "be very vigilant with all resources you use" approach, that includes the whole webdev liftoff, where we ship a few hundred mb chromium engine for a dock app
We'd have far fewer apps, far fewer features, far more bugs, far more crashes, far less stability and far more memory safety vulnerabilities. Oh, and Linux and Mac would be far less usable.
The age of performance optimizations was the age of computers as little islands that didn't need to communicate with anybody or anything, and definitely not outside of a homogeneous LAN environment. It was the age of people having just one device, running one OS, with no expectation of data synchronization. Sharing files was, at best, done by sending quarterly_report_v14_approved_by_legal_fixed.doc over email. This is no longer the age we live in.
In the timeline I remember, Microsoft and Windows were routinely criticized for producing bloated and buggy systems. Especially from those who previously used an Amiga or Mac. A new version of Windows inevitably meant buying a whole new computer, along with upgrading the memory midway through it's 3-4 year lifespan.
I thought we used VC++ 6 to write Abomination[1] but I guess it was VC++ 4 as the dates don't align. The serial cable debugger feature was a godsend. Had the game running on one PC and stepping through it on the other. VSS would kill us, though.
I could've sworn VC++6 had Intellisense. I'm not going to dig too far to confirm but Wikipedia seems to agree with me (https://en.wikipedia.org/wiki/Code_completion#Visual_Studio) - it's not a great reference but definitely implies that it was there:
> the Visual Basic versions of IntelliSense were always more robust and complete than the 5.0 and 6.0 (97 and 98 in the Visual Studio naming sequence) versions of Visual C++
I'm also pretty certain it did, and actually that VC++5 (that I'd used as a student) didn't. I'm less confident about go to definition and stuff, but I think that was all part and parcel of the same thing, so VC++6 had it and VC++5 didn't.
(Just like today, it would sometimes fail to work, for no obvious reason, and it was impossible to figure out why, so perhaps for this article it had got itself into one of those situations...)
Before the .Net era, there were millions of programmers who were experts in VB. In fact, VB6 was the defacto tool to build desktop apps.
Then Microsoft decided to compete with the new-age rivals: Java and CORBA. So it expanded COM into DCOM and then further into COM+, and eventually released the .Net platform.
Suddenly, those millions of programmers and their built desktop apps were obsolete, as they had to race to understand .Net and learn how to use it to build new apps and replacements for the old VB6 apps.
And somewhere along the way, many of them decided it wasn't worth the struggle (because .Net was a nightmare to install as client apps on Windows machines; even the deployment scripts had becom3 too complex), and they migrated to other tools (Java, Python, Perl, Ruby on Rails, PHP, etc.) or to non-programming jobs (usually management).
Thus, within a few years, Microsoft had veritably killed the programming industry it took decades to build and nurture (and yes, Microsoft's decision to turn a blind eye - as its Windows OSes, MS Office and Visual Studio (VB & VC++) tools were pirated across the world, churning out millions of programmers and users familiar with its products as they used the pirated versions at school, college. home and office - that was also a deliberate decision by Microsoft during this halycon era).
But I feel .Net became too big of a beast even for mighty Microsoft to handle. As concerns grew over the performance aspects and innumerable dependencies of the .Net platform and related tools (Azure, SSIS, SSRS, etc.), the world started to shift away from Microsoft's tools, and that's perhaps why Microsoft finally knuckled under and embraced the open-source ecosystem it had openly hated for decades. VSCode, etc., are Microsoft's last-ditch attempts to have some relevancy in the programming industry.
.net was fine ... they were solving these fleeting problems of interoperability, event driven gui programming, object re-use and a bunch of other things. They tried tackling this so many ways: win16, ole, mfc, activex, win32s ... it was a big mess and nothing really worked well.
Microsoft had some really smart people working on the problem for years and .net was the culmination of the efforts with things like c# and the very interesting f#.
The problem was they finally solved the desktop interoperability problem when it no longer mattered and there wasn't a huge killer app for it.
Properly scoped well designed abstractions can be extremely powerful and also pretty useless.
There's an interesting counterfactual if they had .net ready to go around windows 98 ... I might be on a windows phone right now...
Increasing the availability doesn't mean decreasing the price ... people think those are intrinsically related - not so much.
You can get a prada shirt for $2,000 ... as many as you'd like, for $2,000 a piece. No problem. They'll make the factories go burr all night long. Still $2,000.sweeping
There's a bunch of things like this. $100 bills for instance ...
a new entrant might yield a price drop, or, it might not.
> why not? i'm sure they can jump into the hustle.
Not so quick. Critical difference is the relationship between enterprises and the state. In China, the state owns the enterprise, in one way or another. High costs of memory is a threat to the established Chinese electronics manufacturers. The Chinese state can optimize returns at a higher level than the one some petty chip manufacturer operates at, especially if doing so means it could gain coercive geopolitical strength, aka blackmailing.
I mean kind of? Industry and politics are always cohorts. China has structural differences but new entrance to commodities and defensives are almost exclusively price sensitive offers. If say, Singapore or even a firm in Ohio tried to enter the market for global play, they'd undercut as a strategy
Except... Maybe they don't need to. Demand is outstripping supply right now. Competing on availability may be sufficient
You can easily see this on a mini-macro scale with popular restaurants. Often a restaurant with an equivalent menu and prices will open nearly adjacent to a very popular place and can sustain itself simply because you don't have to wait an unacceptable length of time to be seated
I really don't know why people down vote me. What do I need to say to make things for free that people like? Sincere question. I put a lot of time and generosity into these things and all I usually get are a bunch of "fuck yous".
This is honestly an existential issue for me. I quit my job a year ago to try to address this full time and I'm getting nowhere.
It is kind of noisy because the release recency, which is what your "age" column actually represents, is not important data for the comparison you are trying to make.
Also what message we should get from that table is not really obvious.
> I know artificial analysis quite well as the gold standard in llm evals.
I also know them, but it took me a while to realise you were publishing their data in that table. I don't think it was clear.
> The age is important because new techniques keep being developed and so it is a very rough indicator of the size/cost/efficiency trade-off.
Yes but you are already including the name of the model, your potential public for the table already know about model's release history and therefore each model's age, at least roughly.
The json on the page has a coding index result it hides from the table.
That's what this exposes. It's a sorting from the leading evals company on the coding index for basically every model that matters presented in an easy to parse format that you can feed into model routing harnesses in real time so, for instance, your agents can dynamically upgrade themselves to better models as they come out or cost optimize based on eval results.
I do stuff like this, give it away for free and it's either ignored or makes people angry...
I really wish I didn't piss people off with my sincerity but somehow it always goes down that way
reply