Recently I need to reorder the pre-req installation for a InstallShield installer package that has pre-reqs and MSI inside it. That means I’ll have to extract all the pre-req EXE and MSI files and write batch files to call them, then the installer again for the core software, all done silently.
I learned:
Administrative install (/a switch) only extracts the core firmware, not the pre-req files
/b”<target-folder>” extracts the pre-reqs as .prq files
Need to steal the extracted .exe files by monitoring temporary folder as they go
/s doesn’t always silent installs. Some with MSI packed inside the .exe files requires the ‘silent’ request passed to the MSI by adding /v/qn (/v passes the switch to the MSI, which is /qn, which means silent in this case)
.NET before v2.0 requires different switches to install without prompting for license agreement, namely
I was asked to install McAfee Enterprise Antivirus v8.8 as a favor to somebody else’s client, turns out this piece of 5h1t is a nightmare.
After I installed the agent, the main software won’t proceed to install. I looked up error code 1603 in the MSI log and upgraded a fresh Windows XP to SP3 and updated Windows Installer to 4.5, set the time back (somebody suggested potential certificate problem) and it still fails.
Without being compensated for my time, I gave up and tried to uninstall it through Add/Remove programs. Now the uninstaller complained that the program cannot be uninstalled in managed mode. Normally, I’m quite understanding towards programmers since their job is wrestling complexities, but this time there is no excuse: if you add an item in Add/Remove programs, people are expected to be able to uninstall it directly from there by definition!
After some digging, being unable to uninstall McAfee is a known common problem. The IT-crowd at University of Oregon has written an article with the solution. It boils down to running:
I bet even stoners are more competent packaging the software for release! How much productivity across the world they have drained by shipping out an incompetent software bad as malware!
Stay away from McAfee. Won’t install it even if they pay me $8k. Refuse to service anything McAfee without charging an exorbitant sum, or you won’t make your lost time back.
I was trying to image a mSATA SSD with Ghost in Windows (I’m using version 12.0.0.8023 Corporate Edition as I bought the Altiris license) and I ran into internal consistency error 8027 right at the very beginning of the copying process.
For some reason, it doesn’t happen if I boot to the DOS version (provided by Agilent) to do the cloning.
Luckily the status bar tells me what’s going on during the process. I notice it always hangs when ghost tells me that it’s TRIMming the SSD. I looked up the help file (ghost32.exe -help) and noticed that there’s a “-NOTRIM” option. Tried it and the clone completed successfully.
If you live in a mixed environment of Windows 7 and 8/10 computers, you might want to set the usernames to be the same so you can share the files/printers without managing Homegroup.
Nonetheless, in Windows 7/2008, if you try to create a user account using the traditional tool (Local Users and Groups) in Computer Management, you are not allowed to use email address as user name because they banned the at-mark (@):
I searched the web for quite a while, came across stuff like UPN (User Principal Name) without luck (No active directory on Windows 7, nor I want to setup a domain controller in Windows 2008 for home network). Turns out the solution is dead simple: use the “User Account” from Control panel to create the user account. No questions asked!
At work I came across a piece of fine software (fine as the ‘F’ in RTFM) to access the data from an open-source project that decided to throw a warning dialog box every time when it come across some minor data integrity issues while going through tens of thousands of entries.
Upon inspecting the C code, they failed to provide a mechanism to suppress the warning dialog box for Windows (it can be done a function that uses with signals in linux though). Since this is a one-shot gig anyway and I don’t have the time to figure out the dependencies and recompile their code chain, I came up with something ghetto: automated mouse clicker.
It’s the kind of solution that is not too natural to come by for people with finer control over computers (aka, software engineers), but immediately obvious to middle-school level script kiddies.
I have to admit I almost lost the ‘street’ knack as it took me an hour digging through the documentation and code base to find a ‘proper’ way to do it before I even gave the ghetto patch a thought. That’s how seasoned engineers can get pwned by noobs who ‘don’t know a damn thing’ but ‘gets the job done’. Luckily, I remembered to switch hats early enough this time.
A ghetto trick or two can be used ONLY when we are really sure that it’s a one-time gig that won’t be reused. Any attempts to extend the one-time-wonder can only attribute to pain and misery for everybody, as a recurring theme in software and product development projects.
Sounds like a familiar situation at work? It’s always that one small thing your client asked you for. If you hack something up quick-‘n’-dirty and it worked, they are going to say, ‘by the way, one more one small thing…’. Lather, rinse, and repeat and soon you’re buried in brown, sticky foam.
This can be explained by the life lesson I learned in my first summer internship during high school: people in general don’t know what they really wanted and they don’t communicate well. Me included. The soft-skills coaching received at Stanford helped quite a bit, but I still consider myself ‘not there yet’.
Typically, your client is going to ask you for something specific (which is not what they really wanted because they haven’t figured it out yet) and keep revising and making sharp turns at your expense to make up for the ‘errors’ as in ‘trial-and-error’.
This is not a win-win scenario. It’s a tug of war between you and your client: you want your client to do their homework to figure out what they wanted and your client want to crap-shoot at your expense. It creates tension between you and your client. A work contract (or a specification agreed upon) might protect you from blame, but a client that doesn’t get what he/she wanted (regardless of who is at fault) is not going to be happy about your work.
One way out of it is to ask your client to start with a vague, underlying theme of what they are trying to accomplish. This way they won’t be overwhelmed by the pressure to get all the detailed specs right. If you have a long term relationship with the client, it pays to learn what are their general ‘interests’ are so you can piece the fragments together quicker. Gradually, you can ease them into thinking through their underlying goals aloud and you can take the opportunity to explore how you can help them accomplish it.
Guide them through their thought process, and maybe show a little prototype or design to help them visualize how they can accomplish their goals (through you). With the big picture in mind, you can make educated judgement, fill in details, and decide what intermediate/iterative steps are needed instead of letting the client unwittingly micromanage you based off his/her solution approach by saying ‘I need one small thing quick…’ one at a time. You are now in control over the project and in a very good position to help your client meet his/her true agenda.
It seemed like more upfront work to meet your client halfway and do part (or half) of their homework, but for anything other than a short hit-and-run, it’s less total work for you and a happier outcome for both. Using this approach, I was able to deliver a last minute change in ‘1 minute’ because the software architecture I designed was tailored towards the client’s underlying goal, so the feature he forgot to ask for was naturally built in and only requires one line of code to turn it on despite I hadn’t explicitly planned for it.
The bottom line is that humans are clueless (at various levels) and they want to further their interests (by doing something). If we take both parties’ interest into account and spot inefficiencies that we can correct, there’s at least an approach where everybody wins. Game-theoretic approach and market equilibrium is ‘optimal’ only when ‘technology’ is held unimproved and there is no on-going relationship between parties. Life doesn’t always have to be a war or struggle 🙂