So one of my friends finally upgraded his computer (good stuff, I was thinking into stealing some of the components for my personal usage ;D), and dropt me a call to install the OS, and make some post-assembly work (moving his old hard drive there basically).

The thing is that, while he installed Vista in the main HDD, he wanted to keep XP in the old one. Since the original installation was beat up, and new components would not do any benefit to it, I simply got the HDD there, and left it disconnected. Besides, Vista complained about the new HDD not being the main one, and thus the OS installation phase was halted at that point.

So I installed Vista, changed something here and there, while admiring Vista idiosyncrasies and hunting for moved options. Also, it seems now you need an antivirus if you want to fully "enjoy" WLM chat facilities (and they wonder why I don't use it...). Everything ran fine, though, so I copied the old files to the new drive, turned off the computer, and proceeded to connect the older one and install XP. Just like before, everything went fine, and with my job nearly done, I went to Google for fixing up possible bootloader issues.

You see, apc published a few tutorials of how to dual boot between various OS, including Vista and XP, so I followed their tips to the letter. The problem is that: first, unlike they say there, XP didn't mess with Vista bootloader, and second, EasyBCD did not find XP in the other drive, so here I just opted to fix such thing myself with the tool, because in theory it should work.

OK, it didn't. XP did not start in any way, but fortunately Vista remains fine. I did some googling, and found another way to do it, old-schol, but reliable, using bcdedit. According to the guy who wrote the guide, it should work with two separate drives, but did not again.

So, the question is how to do it correctly. In my mind, a few explanations for the failure cooked over, like that the old drive was disconnected when I installed Vista, and he does not like it now, or that because of the peculiarities of the motherboard southbridge (SB700), everything is running in IDE mode (I nearly panicked when I peeked into the BIOS, and did not see SATA in any place), and the fact that the old drive is in channel 0 as the master device, and both the optical drive and the new HDD are in the channel 3 as slave and master respectively (weirdness) causes them to fight over bus dominance or something. Or perhaps that I should get some sleep, because I'm making up extravagant explanations with no basis in reality, and missed something very evident.

So, I'll appreciate if someone did something like this before would tell me how they did it, or where did they find the information about how it should be done.

Yes, you're reading it right. No, nobody took over my account. I'm here, complaining about the Opera browser, and specially about the latest version with which Opera Software has achieved the impossible: Having an Internet-compatible browser (mostly) after years of trying and web creators constantly giving them cold shoulders. Yet, they also done another incredible thing: to completely undermine all this effort due to a half-assed new feature introduced in Opera 9.5 named content indexing.

As you may know, the latest trend when it comes to web browsing history is to index not only the URLs you visit, but also the page title, and letting you search every single character in them directly from the URL bar. The boon to procrastination productivity is immense, yet Opera wished to go a step further: to index ALL text contained in every single page you visit. The idea is even more brilliant, and while it turned out it's not as useful as having URL searching, when you need it, it's a wonderful feature.

If it worked.

Let me explain. Turns out when they devised the frontend for it, either they smoked something or simply did not care. Instead of doing something so simple, like, well, list FIRST the results of the URL search, and later the results of content search, they mixed both, so you end up with tons of noise in the address bar and effectively killing much of its usefulness as you have to pay attention to the underlined results to actually found what you really want. An even better way to implement this frontend would be putting up the content search part under an alias (there's actually one), so the user could launch it on demand, and leave URL search alone in the address bar.

The problem gets worse when you see that, for no good reason, the program simply refuses to index certain pages titles, just the URL addresses. And that the number of results listed by the feature is fixed, and sometimes what you want does not appear in the results list, despite knowing it's there.

It should be said that those problems appear as time passes and Opera's history gets filled with items. Precisely this, coupled with my desire to have a page history as big as possible, gets me to the ultimate weak point of this feature: performance. You see, the search function is pretty snappy as long as the word actually exists in the history file. But sometimes it turns out it's not in there, and it's then when you witness the self-proclaimed "Fastest browser on Earth" bringing modern HDDs to their knees during sometimes as much as 20 seconds.

I constantly experiment this due to my appreciation for the possibility of searching things directly from the address bar, and it's something I use several times a day. But while a single programmer as mouser implemented an option to stop searches when the inputted string matches an alias, a company with more than 500 employees wasn't able to do the same, and so I'm forced to constantly witness how Opera freezes itself and the drive works like mad trying to find something a string like "g cheese" (which launches a search in Google for cheese). Of course, there's a way to fix this and it's typing the string and waiting until the drive stops thrashing around (hooray for HDD leds), and only then you press "Enter". With this trick, you can actually use the browser, but if you input the string and press Enter, prepare yourself to see how even the whole computer gets nearly unusable, and I'm running a dualcore machine. Should I get a RAID 0 array and a quadcore to "fix" Opera Software errors?

There's actually a way to end with all of this. But the problem is that you get reverted to pre-Opera 9.5 behaviour (save for the fact that you don't have to type "http://www." to find every address). And this is either deactivating content indexing, or going into opera:config and disabling content search from the address bar. Since both URL addresses and page titles are stored in the "opera.dir" file, I thought that I could replicate Firefox 3 behaviour while keeping Opera advantages and using them as needed. Turns out I was wrong.

You see, the only way to reach Codex Transportica, for example, is typing "codextransportica" or "codex", not "Codex Transportica", nor "Transportica", despite seeing the "Codex" part of the title bolded as you type it. A more complex page like the Recent Posts section of this very forum is unreachable by typing "Recent Posts Donation Coder", while in Firefox 3 is totally possible. So, you either bring back content indexing, or you get forced to remember entire addresses. Which, frankly, I prefer considering the number of I/O operations done by the browser to update indexes which I rarely use, and the aforementioned noise.


This is perhaps the most glaring issue with Opera post-9.5. The other one that troubles me is the way memory gets released, which makes that after a while without touching a single page, switching the focus back to its tab shows you a little slide show as you see the browser repainting the entire page, and reloading the pictures, probably from the cache. And then, minor issues, like the overflowing loading bar when the page loaded has more than 100 elements, artifacts in the history panel, or dialogs that continue to be in English. Practically everything mentioned here has been reported on the official bugtracker by me, and there are several threads about them in the Opera forums. But they continue to be unfixed, and probably they won't be until Opera 10.

I'm used to navigate around software shortcomings, and bugs, but this is a joke. No wonder some Opera users complain as loud as they do, even if the company chooses to ignore them. I am starting to believe what I read here a while ago (I think Josh wrote it), that Opera do things as they want and they rarely finish the work. Perhaps if they packed less features in the program (let's be sincere, there are tons of features included in Opera just for the sake of being included or to please a really small group of users, scroll marker coming to mind, for example), and they focused on the really useful ones, perhaps they could have avoided the train wreck that Opera 9.5x was (Opera 9.6x, while not perfect, is much improved in all areas).