Frankly the tools really sucked back then. Build systems and IDEs were awful. Just go play with an autotools-based C project sometime to remind/education yourself. Visual Studio 7, horrible. No mainstream refactoring IDEs to speak of. C++ compilers across multiple platforms were horrible at standards compliance consistency, and you could barely get a working STL, practically everyone wrote their own string and containers classes. CORBA -- some nice ideas, bad in practice. Java was a dumpster fire of EJB/J2EE heavyweight, with slow an d expensive application servers. Expensive Oracle installs dominated the database world, with the rest shored up by MySQL installs that were only partially ACID. No CSS HTML, pre-HTML5 so a mess of nested tables to make things lay out properly. Most sites were a pile of spaghetti code "type 1" JSPs or ASPs or really bad PHP sites making database calls and queries right in the page source, horrible to maintain.
Then the serving or hardware infrastructure, in the world of web stuff... forget about cloud or even reasonably priced hosting services. Most shops, even small ones, I worked at ended up having their own sysadmin team managing an owned or rented fleet of expensive Sun server hardware, etc. Closets full of hot and pricey hardware etc.
And as for languages... I learned Python in 95 or 96, back when it was pretty new. But almost no shop would have considered hiring me to work in it. Erlang, Python, OCaml, various Lisps, Smalltalk, all that good stuff all _existed_ but pretty much nobody would ever consider letting you write production code in such "weird stuff" until Ruby kinda broke the barrier. Perl was everywhere, but "serious" shops started to push Java, but Java was frankly awful back then around 2000. As I allude to above C++ was painful to work in at the time. C# didn't really exist yet. Visual Basic was all over the place, but was frowned on for "serious" stuff.
I think people forget how dominant and awful "enterprise" development is/was. It's still out there, but HN in general doesn't seem as exposed to it. Back in the late 90s, early 2000s, the accepted "enterprise" stack was the aspirational crap _so_ many shops adopted... it was for that time what "microservices" and "bigdata" other dogma are today. People didn't need it, but they thought they did.
Frankly, everything took longer to get done. Simple things are quicker to get done now.
Nah, it wasn't a particularly good time to be doing software dev.
I guess if you were employed in the right place, and were lucky, you would at least get to work on pioneering work building the tools and infrastructure that we now take for granted and complain about. Being at a Google building Bigtable etc. or Sun Microsystems working on the innards of Java etc. back then would have been a dream job. But the vast majority of us never got that chance. We were plumbers, too, just with really crappy pipes.
> Just go play with an autotools-based C project sometime to remind/education yourself.
Funny you should mention autotools. It reminds me a lot of webpack, especially in the way nobody[0] really understands how it works, but you search around for examples and copy/paste what works for you.
[0] a hyperbole; i'm sure someone does, just as I'm sure some people dreamt of M4 macros back in the day
Oh it's so awful; but I recently was exposed to a project where some of the leads were trying to defend it as a reasonable tech choice. It really isn't. Not in this era. Most people never even used it correctly in the first place.
Totally agree, and let's also mention Stack Overflow and the plentiful learning resources on Youtube etc. In 2000, I was a teenager trying to learn C++ and the Win32 API, and when I got stuck I got really stuck. These days, the amount of resources to help you with a problem or learn a new technology are infinitely higher which removes one of the most frustrating aspect of software dev.
It wasn't autotools per se that was horrible. It was the the fact that you had to pollute your system with random libraries, often no longer available from the operating system vendor repositories. Docker has been a lifesaver with these older projects.
Frankly the tools really sucked back then. Build systems and IDEs were awful. Just go play with an autotools-based C project sometime to remind/education yourself. Visual Studio 7, horrible. No mainstream refactoring IDEs to speak of. C++ compilers across multiple platforms were horrible at standards compliance consistency, and you could barely get a working STL, practically everyone wrote their own string and containers classes. CORBA -- some nice ideas, bad in practice. Java was a dumpster fire of EJB/J2EE heavyweight, with slow an d expensive application servers. Expensive Oracle installs dominated the database world, with the rest shored up by MySQL installs that were only partially ACID. No CSS HTML, pre-HTML5 so a mess of nested tables to make things lay out properly. Most sites were a pile of spaghetti code "type 1" JSPs or ASPs or really bad PHP sites making database calls and queries right in the page source, horrible to maintain.
Then the serving or hardware infrastructure, in the world of web stuff... forget about cloud or even reasonably priced hosting services. Most shops, even small ones, I worked at ended up having their own sysadmin team managing an owned or rented fleet of expensive Sun server hardware, etc. Closets full of hot and pricey hardware etc.
And as for languages... I learned Python in 95 or 96, back when it was pretty new. But almost no shop would have considered hiring me to work in it. Erlang, Python, OCaml, various Lisps, Smalltalk, all that good stuff all _existed_ but pretty much nobody would ever consider letting you write production code in such "weird stuff" until Ruby kinda broke the barrier. Perl was everywhere, but "serious" shops started to push Java, but Java was frankly awful back then around 2000. As I allude to above C++ was painful to work in at the time. C# didn't really exist yet. Visual Basic was all over the place, but was frowned on for "serious" stuff.
I think people forget how dominant and awful "enterprise" development is/was. It's still out there, but HN in general doesn't seem as exposed to it. Back in the late 90s, early 2000s, the accepted "enterprise" stack was the aspirational crap _so_ many shops adopted... it was for that time what "microservices" and "bigdata" other dogma are today. People didn't need it, but they thought they did.
Frankly, everything took longer to get done. Simple things are quicker to get done now.
Nah, it wasn't a particularly good time to be doing software dev.
I guess if you were employed in the right place, and were lucky, you would at least get to work on pioneering work building the tools and infrastructure that we now take for granted and complain about. Being at a Google building Bigtable etc. or Sun Microsystems working on the innards of Java etc. back then would have been a dream job. But the vast majority of us never got that chance. We were plumbers, too, just with really crappy pipes.