I know a lot of the people who maintain XLW are frequent contributors here, so I am hoping this is the best place to ask this question.I've been using XLW for a long time and it works beautifully (version 3.0.0f). However, I encountered a memory-leak problem, which only shows up intermittently. This particular function uses ADO to run a query against a SQL Server and returns the results in the spreadsheet. For certain queries, it shows a heap allocation error. In debug mode, it breaks in XlfExcel::FreeMemory, on the line "delete[] freeList_.back().start;". It doesn't seem like it does anything wrong. I can even ignore the error (when the dialog with "Abort|Retry|Ignore" shows up), and everything keeps working as it's supposed to. The particular query exhibiting the behavior doesn't even return that much data, it's about 300 rows and 3 columns. For that size CellMatrix, XlfExcel::GetMemory is requesting in the neighborhood of 60K bytes. If I change the initial buffer size to 256K, the error disappears. The other thing I discovered is that it seems related to how I am reading strings returned by ADO. The ADO type for those is adVarChar, which is represented as a BSTR, which is typedef'ed as a wchar_t *. If I just return a constant string instead of converting the BSTR to a std::string (using WideCharToMultiByte), the problem goes away as well.I guess my questions are the following:- has anyone experienced any similar behavior? How did you solve it?- would switching over to version 4 get rid of this problem?- is the solution I came up with (increasing the initial buffer size) a real solution or will it just make the problem appear less often?Thanks,Marc.

I did see something similar while i was looking at some multi-threading issues, there did seem to be a couple of issues with the memory allocation code in versions 3&4. Probably you best bet is to try the latest release from the dev5 branch and see if it reoccurs. Are there any multi-threaded functions in your xll? If not it may well be to do with buffer size as you've found, i think that your fix will be ok in the very short term but yes it is probably just hiding the issue with your current memory load. John

Nope, nothing multi-threaded in my code. It's really strange because this problem only manifests itself very rarely, I think it's only happened twice in the past 2 years.

Hmmm, it may be related to the string functions then, there were a few issues in there, also fixed in the Dev5 code. Would it be possible to test with that?John

I wish I could tell you I will definitely test it, but unfortunately I might not get to it for a while. I will most likely try it though if the problem happens again. Thanks.