mionica

That error can be also generated by attempting to integrate a broken archive (it happened when I attempted to integrate a malformed Ryan's UP - somehow my download got corrupted). Try extracting the hotfix/update pack/whatever manually to make sure (just in order to check its consistency). If the archive is broken, just download it again. It happened to me yesterday evening EET, on a corrupt RyanVM Update Pack 2.1.7; upon downloading again, it worked. PS. Also regarding latest RVMUP: you need set nLite to prompt on integration, and choose to overwrite some (apparently) newer files, for example mfc42u.dll - otherwise you won't get correct integration.

@nuhi Sorry for my "tone" - I didn't mean to offend you, or anything (and I couldn't want to, since I - like many others - owe you a great debt of gratitude for creating nLite, which I use every month at least 2 times ). Try to imagine me as expansive (which I am), or childish (same) instead of disprespectful (which I am not, at least not towards you). And the other thing I asked clarification on? What's the foder size CAB entry, or if it would take too long to explain, where can I find the appropriate informative resources?

@nuhi just in which OS versions does that happen? I have a WinXP SP2, which I manually converted to VLK (changed the setupp.ini$[Pid]$Pid value); I got it by integrating SP2 over the RTM (using the good old /integrate switch of the SP). My driver.cab is dated august 23, 2001 and it does not contain tcpip.sys. Also, my sp2.cab is dated august 3, 2004, and it also doesn't contain tcpip.sys. The t*.sys files I have are (alphabetic listing) - in driver.cab: t2r4mini.sys, tandqic.sys, tbatm155.sys, tdk100b.sys, ... ; - in sp2.cab: termdd.sys, ... Could you at least check if CABs rebuilding is necessary? -- And one more thing. You said Could you develop a bit further, or post a link which does so? I'm afraid I didn't get the picture.

@nuhi Another thing, regarding drivers. I assume it would be possible to leave the driver .cab's unaltered at all times. For instance, if an addon (RVM Update Pack, for instance) would need to replace a file (say, ntoskrnl.exe) belonging to a driver .cab, wouldn't it be much faster to just copy the new file to \i386, pack it, and remove its entry from drvindex.inf? I'm currently testing if this would work. But it makes sense, since dosnet.inf tells setup anyway which files should be copied, regardless of their location (in \i386 or in the .cab's). I, for one, would greatly enjoy not having to wait for ages for driver recompression to take place. It would be great if nLite would give me this option. This is my first point. Further reasoning would go like this: - if drivers recompression occurs, with compression set to LZX:21, setup gets extremely slow (and cpu-intensive) during textmode setup - particularily annoying during VM test sessions - if drivers recompression occurs, with default compression (MSZIP?), the resulting .cab's are much larger than the initial ones. Consider that once makecab'ed with LZX:21, all files needing update until now in both driver.cab and sp2.cab (I'm talking RyanVM's UP now) take up only 5.48MB, which is by far less that the overhead the default compression takes. Regards.

@nuhi it occured to me that, at least in the version of xp/sp2 i have, tcpip.sys does not reside in the driver .cab's, instead it is outside of them. would it be complicated to trigger driver recompression only if tcpip.sys resides in one of those .cab's? however, if there is another reason for triggering recompression, could you tell it to me?